# SWE-2 Flash > SWE-2 Flash is Cognition's small, fast coding model, mid-trained and reinforcement-trained from Thinking Machines' Inkling-Small (276B total, 12B active parameters). It runs inside Devin Desktop and Devin CLI, with no model API of its own and no public weights. - Canonical: https://www.anchorterminal.com/tools/swe-2-flash - Pricing: https://www.anchorterminal.com/tools/swe-2-flash#pricing - Reviews: https://www.anchorterminal.com/tools/swe-2-flash#reviews - Markdown: https://www.anchorterminal.com/tools/swe-2-flash.md (~9,600 tokens) - Slim: https://www.anchorterminal.com/tools/swe-2-flash.min.md (~1,830 tokens, same facts, less prose, for token-sensitive contexts) - JSON: https://www.anchorterminal.com/tools/swe-2-flash.json (this page as data, same URL with Accept: application/json) - Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt) - API: https://www.anchorterminal.com/api/v1/index.json - MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key) - DNS: dig +short TXT swe-2-flash.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.. and .vs.. under the zone, and the same under grade.letme.dev) - Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data) - Updated: 2026-11-04 ## Overview **Grade F · 36.4/100 · rank #1243 of 1350 · #33 in Model APIs & inference · not agent-ready · confidence low** ## Assessment A new small coding model from Cognition that anyone can try free in Devin Desktop and Devin CLI until 30 November 2026, with strong vendor-reported coding scores. It has no model API, no published per-token price, context length or model card, and no public weights, so an agent can reach it only by driving Devin's own CLI. ## Facts | Field | Value | | --- | --- | | Vendor | Cognition AI, Inc. (https://cognition.com/blog/swe-2-flash) | | Kind | Model API | | Category | Model APIs & inference (https://www.anchorterminal.com/categories/inference) | | Auth | OAuth or key · No model API or key of its own. A person signs in to Devin Desktop, or runs `devin auth login` for Devin CLI, through a browser (enterprise users through their identity provider). The CLI then keeps an API token in credentials.toml that by default is persistent and does not expire, and the docs warn that anyone with the file can authenticate as that user. Enterprise admins choose which models a team may use with a model allowlist. | | Pricing | Freemium (Freemium) · Free in Devin Desktop and Devin CLI through 30 November 2026, per the launch post. Afterwards it is billed against the plan's daily and weekly quota, then extra usage at API pricing. The post says one-fifth the price of SWE-2, whose list price is $3.00 input, $0.30 cached input and $15.00 output per million tokens, but no SWE-2 Flash figure is published: the docs' cost table loads in the browser and the models pages don't list it yet (checked 2026-11-04). Plans are Free $0, Pro $20 a month, Max $200 a month and Teams $80 a month plus $40 per full seat. The docs say Free tier users only have SWE-1.6, so whether Free reaches SWE-2 Flash is unconfirmed. | | x402 | No · No model endpoint to pay. No x402, MPP or L402 on the Devin pricing page, the Devin Desktop models page or the launch post (checked 2026-11-04). | | Licence | Proprietary model, used through Devin Desktop and Devin CLI under Cognition's Platform Terms of Service. The weights are not public; Cognition says it is building a programme to share them with industry partners and researchers | | Docs | https://docs.devin.ai/desktop/models | | llms.txt | https://docs.devin.ai/llms.txt | | Last release | 2026-11-04 | | Surface graded | The model as Devin Desktop and Devin CLI expose it, read from the launch post, the Devin docs, the pricing page, Cognition's legal pages and the status page on 4 November 2026. No account was made and nothing was run. There is no model API to grade. Devin itself is the separate devin listing | | Model | Mixture of experts, 276B total and 12B active parameters, the size of its base model, Inkling-Small from Thinking Machines Lab (the inkling-small listing). Cognition added a mid-training stage of a few trillion tokens of scientific articles, blogs, code and other sources in three stages, then large-scale reinforcement learning. Context length, output limit, modalities and reasoning levels aren't stated | | Where it runs | Devin Desktop and Devin CLI, per the launch post. Not in the Devin v3 API, whose session agents are `swe-2-medium`, `swe-2-high` and `swe-2-max`, and not yet named on the Devin Desktop or CLI models pages, which list SWE-2, SWE-1.7, SWE-1.7 Lightning, SWE-1.6 and SWE-1.6 Fast | | Weights | Not public. Cognition says it wants enterprises, governments and researchers to be able to host, fine-tune and study the model, but is holding the weights back because the cyber training that raised ExploitBench to 34.8 per cent could also help attackers. It is building a programme to share them with trusted industry partners and research labs, with more 'in a few weeks' | | Cognition's figures | Vendor-reported. Terminal-Bench 4.0 20.0, DeepSWE 1.1 60.1, FrontierCode 2.0 24.0, FrontierCode 1.1 Main 45.5, Terminal-Bench 2.1 89.2, SWE-Atlas QA 50.0, SWE-Bench Pro v2 Hard 89.9 and ExploitBench 34.8, against Inkling-Small, Beam 1.0, Muse Spark Open, GLM 5.3 Flash, DeepSeek V4.1 Flash, GPT-6 Luna, Claude Haiku 5.5 and SWE-2. 200 tokens a second against SWE-2's 60, shown as a simulated replay | | Pricing | Free in Devin Desktop and CLI through 30 November 2026. 'One-fifth the price' of SWE-2 (list $3.00 input, $0.30 cached input, $15.00 output per million tokens), with no SWE-2 Flash figure published. Plan usage is a daily and weekly quota, then extra usage at API pricing | | Data use | Cognition's platform terms of 30 June 2026 let it use customer data for model training unless a paid customer opts out. Enterprise data isn't trained on without written consent (as read for the devin listing) | | Status | www.devinstatus.com (Atlassian Statuspage) has Desktop Agent and Desktop Tab components alongside the cloud ones. In the 90 days to 4 November 2026 its feed lists 'Devin unavailable' for 20 minutes on 13 August with no component named, Devin Cloud unavailable on 13 August, the web app down on 24 September and minor cloud incidents. None names Desktop or the CLI. A scheduled upgrade of Devin Desktop's extension marketplace on 7 August is read as maintenance | | Family | SWE-1 and SWE-1.5 were Windsurf's models; Windsurf is now Devin Desktop, and Cognition publishes the SWE line, SWE-1.5, SWE-1.6, SWE-1.7 and SWE-2 (10 September 2026), on its own blog. None of them is listed | | Capabilities | inference.llm | | Tags | official, hosted, model, closed-source, coding, moe, cli, desktop-app, llms-txt, status-page, soc2, iso27001, sla, freemium, new | | JSON | https://www.anchorterminal.com/api/v1/tools/swe-2-flash.json | ## Score breakdown (methodology v0.7, October 2026 research run) Assessed 2026-11-04 from public evidence against the published checklist (https://www.anchorterminal.com/benchmark/#checklist). Confidence: low. Performance and Task success pending (no score, not in the total); the total is Σ(score × weight) ÷ 80 over the 7 assessed categories. "This run" is each category's share of the 100 points. | Category | Weight | This run | Score (0–100) | Points | | --- | --- | --- | --- | --- | | Reliability | 16% | 20 | 50 | 10.0 | | Performance | 10% | pending | pending | n/a | | Schema & documentation | 13% | 16.2 | 29 | 4.7 | | Agent ergonomics | 13% | 16.2 | 10 | 1.6 | | Security & auth | 14% | 17.5 | 53 | 9.3 | | Payments & pricing | 10% | 12.5 | 20 | 2.5 | | Task success | 10% | pending | pending | n/a | | Maintenance & community | 7% | 8.8 | 53 | 4.6 | | Transparency & trust (editorial 56, provenance 72) | 7% | 8.8 | 64 | 5.6 | | Negative events | up to −15 | up to −15 | The launch post says 'On leading evals, SWE-2 Flash outperforms all American open-weight models'. The post's own table, a first-hand source, gives Meta's Muse Spark Open (American, open weights, released 15 October 2026) 54.0 on SWE-Atlas QA against SWE-2 Flash's 50.0, and Meta's own figure for Muse Spark Open is 58. A 'surpasses' claim a first-hand source shows false is a deduction of 2 (research/new-2026-11-04-ripple-swe2/NEWLIST.md, rule 4). The case against: SWE-2 Flash leads every other row it shares with the American open-weight models in the table, and DeepSWE 1.1 is 60.1 to 60.0, so 'on leading evals' can be read as overall. Proposed for the founder to confirm at landing, as xiaomi-ocr-0's was | -2 | | **Total** | | | | **36.4 → F** | ### Why each score - Reliability 50 (checked 2026-11-04): Model reading, for the surface it runs on, Devin Desktop and Devin CLI. Status page at www.devinstatus.com (Atlassian Statuspage) with Desktop Agent and Desktop Tab components besides the cloud ones, read on 4 November 2026 (20). The history feed for the 90 days from 6 August lists 'Devin unavailable' for 20 minutes on 13 August with no component named, read as across the board and a major as on the devin listing; Devin Cloud unavailable on 13 August and the web app down on 24 September name cloud surfaces this model doesn't use; the rest are degraded-performance incidents (new sessions on 3 September, 'multiple services' on 17 September, Ask Devin on 26 October, automation triggers on 4 November) and session queuing on 10 and 11 August, read as minor and a scheduled Devin Desktop extension marketplace upgrade on 7 August, read as maintenance. One major (10). No rate limit figures for the model or the CLI; usage is a daily and weekly quota with no numbers (0). No 429, backoff or retry guidance for this surface (0). Cognition publishes an SLA (25 February 2026) with uptime and response times that applies when included in an order form, read as a contractual SLA on Enterprise, as on other listings (10). The launch post makes it available today with no preview or beta label (10). 20 + 10 + 0 + 0 + 10 + 10 = 50. - Performance: Pending. Latency is measured per call by our probes, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until the first probe window closes. - Schema & documentation 29 (checked 2026-11-04): Model reading. No model API, so no OpenAPI entry; the Devin v3 OpenAPI file lists SWE-2 agents only, and no model card was found (0). docs.devin.ai serves llms.txt and Markdown copies of every page (10). The launch post says what it is for (automations, code review, feature development, documentation) and the CLI docs say when to pick a SWE model, but the models pages don't list SWE-2 Flash yet (6 of 20). No documented model ID, reasoning settings, context or output limits to type (0 of 15). The CLI pages show `--model` and `/model` examples, with no model-specific errors (3 of 15). Named, dated model versions on Cognition's blog and public release notes, but no release note for SWE-2 Flash yet; the newest is 2 November (10 of 15). 0 + 10 + 6 + 0 + 3 + 10 = 29. - Agent ergonomics 10 (checked 2026-11-04): Model reading (tool use, structured output, caching, context, batch, SDKs). Tool use is Devin's harness, not something a caller configures; no context length, output limit or cache price is published for SWE-2 Flash (5 of 25). No documented output-size or reasoning controls for this model; SWE-2's effort levels aren't stated for Flash (0 of 20). No model-level error documentation (0 of 20). No idempotency or retry guidance (0 of 20). One flag or picker selects it and `devin -p` runs a single turn and prints to stdout, but there is no SDK or direct endpoint (5 of 15). 5 + 0 + 0 + 0 + 5 = 10. - Security & auth 53 (checked 2026-11-04): Read as a hosted model reached through Devin's local clients. The CLI signs in through a browser, or the enterprise identity provider, and stores an API token in credentials.toml that by default is persistent and does not expire and carries the user's access, with no scopes found for it (10 of 30). Devin CLI documents an OS-level sandbox with network filtering, hooks that can block tool calls, and enterprise model allowlists and team settings (14 of 20). Guardrails against prompt injection are Enterprise only and cover user messages, as on the devin listing (8 of 15). Enterprise analytics count tool calls and consumption by model and user; no per-user audit log for the CLI was found (8 of 15). SOC 2 Type II and ISO/IEC 27001:2022, a disclosure address, no bug bounty and no security.txt, per the devin listing (13 of 20). Training: self-serve customer data may train Cognition's models by default, with an opt-out on paid plans that also turns on zero data retention at model providers. Cognition withheld the weights over cyber misuse, which is recorded as a fact, not scored. 10 + 14 + 8 + 8 + 13 = 53. - Payments & pricing 20 (checked 2026-11-04): Published rubric. No payment protocol (0). Plan prices are public, but no per-token price for SWE-2 Flash is on a readable page; the docs' cost table loads in the browser and the post gives only 'one-fifth the price' of SWE-2 (10). Free through 30 November 2026 on Devin Desktop and CLI, but the docs say Free tier users only have SWE-1.6, so whether the Free plan reaches it, and without a card, isn't confirmed; the in-between value used on the devin listing (10). A person signs in through a browser (0). 0 + 10 + 10 + 0 = 20. - Task success: Pending. Task success needs the category task suites run through each tool, which haven't run yet, so this run doesn't score it. Its weight is shared across the assessed categories until then. A data provider's data-quality score is published on its listing now and becomes half of this category when it's scored. - Maintenance & community 53 (checked 2026-11-04): Model reading. Released 4 November 2026 (30). Churn and notice in place of release counts: SWE-1.6, SWE-1.7, SWE-1.7 Lightning, SWE-2 (10 September) and now SWE-2 Flash in a few months, with no model deprecation notice period found and short names that move to the newest model (10 of 20). Public release notes and a support address, no public tracker (10 of 25). No official SDK for the model (0 of 15). The docs are kept current but don't yet cover this model (3 of 10). 30 + 10 + 10 + 0 + 3 = 53. - Transparency & trust 64 (checked 2026-11-04): Editorial half. Closed model under Cognition's platform terms; the weights aren't public and the sharing programme isn't described yet (15). Terms, privacy policy, security page and DPA agree, with training on self-serve data by default and no retention periods, as read for the devin listing (18). The API versioning policy covers API endpoints, not models, and no model deprecation notice was found (5 of 20). Sub-processors listed with purpose and country, per the devin listing's 8 October check (18). 15 + 18 + 5 + 18 = 56. Fix list, everything this grade says the listing lacks, the biggest gain first (21 items). What it covers is at https://www.anchorterminal.com/fixes/swe-2-flash.md (JSON https://www.anchorterminal.com/fixes/swe-2-flash.json). The whole list for a coding agent comes from `POST https://www.anchorterminal.com/api/v1/fix-list` with `{"listing": "swe-2-flash", "plan": "read"}`, free while the beta runs ### What we couldn't check - Deduction of 2 proposed on 'outperforms all American open-weight models on leading evals', against the post's own SWE-Atlas QA row (Muse Spark Open 54.0, SWE-2 Flash 50.0). The founder decides whether 'on leading evals' means each eval or overall; without it the deduction is 0 - Kind. Graded as `model` per the brief, though the only way in is Devin's CLI and Desktop. If the founder prefers, it could be tagged `not-for-agents`, but `devin -p` does let another agent run it single-turn - unchecked: the SWE-2 Flash model ID, context length, price after 30 November and Free plan access. The docs' models pages and cost table don't show it yet; re-read them in a week - unchecked: the weights programme, promised 'in a few weeks'. When the weights reach partners or the public, the open-weight question comes back (a gated or partner-only weights listing, linked by sameModel) - Security programme, sub-processors and data-use readings are taken from the devin listing (same company, re-checked 4 November), not re-read here - SLA line 10 on Cognition's published SLA that applies when an order form includes it. The devin listing has 0 for the same line because it didn't find this page; worth aligning at its next re-check - research/recheck-2026-11-01/RULES-SUMMARY.md, which the brief points to, doesn't exist in this worktree. The fetching rules were taken from the brief's own summary of it - No independent results for SWE-2 Flash were found; artificialanalysis.ai and openrouter.ai weren't requested, per the rules ### Sources - launch post, 4 November 2026, with the benchmark table, training, weights and pricing statements: (seen 2026-11-04) - blog index, listing the post as 2026.11.04: (seen 2026-11-04) - SWE-2 launch post, 10 September 2026, SWE-2's own figures: (seen 2026-11-04) - sitemap, Cognition's SWE posts (swe-1-5 to swe-2) and legal pages: (seen 2026-11-04) - robots.txt: (seen 2026-11-04) - platform terms of service (30 June 2026), read for automated-access limits: (seen 2026-11-04) - website terms of use (13 May 2024), read for automated-access limits: (seen 2026-11-04) - service level agreement (25 February 2026): (seen 2026-11-04) - robots.txt: (seen 2026-11-04) - llms.txt: (seen 2026-11-04) - Devin Desktop docs index: (seen 2026-11-04) - Devin Desktop models, SWE family and SWE-2 pricing notes: (seen 2026-11-04) - Devin CLI models, --model and /model: (seen 2026-11-04) - Devin CLI essential commands, devin -p: (seen 2026-11-04) - Devin CLI auth, credentials.toml: (seen 2026-11-04) - release notes 2026, newest 2 November: (seen 2026-11-04) - Devin v3 OpenAPI file, session agent enums: (seen 2026-11-04) - robots.txt: (seen 2026-11-04) - pricing: (seen 2026-11-04) - robots.txt: (seen 2026-11-04) - status page and components: (seen 2026-11-04) - status history feed: (seen 2026-11-04) - security, disclosure address and data use (as read for the devin listing): (seen 2026-11-04) - data processing statement and sub-processors (as read for the devin listing): (seen 2026-10-08) - trust centre (as read for the devin listing): (seen 2026-10-08) ## Score history Every move of a score since the listing was graded, newest first, with its cause. A score can't move without an entry here. JSON: https://www.anchorterminal.com/api/v1/tools/swe-2-flash/history.json - 2026-11-04: graded 36.4 F (methodology 0.7) ## Who's behind it (provenance 72/100, checked 2026-11-04) | Check | Finding | Points | | --- | --- | --- | | Legal entity named | Cognition AI, Inc. | 20/20 | | Domain age | cognition.com, no registry record we could read | 0/15 | | Endpoint on the vendor's domain | cognition.com | 15/15 | | Terms of service | read, states 6 of the 7 things a reader expects, and has 1 clause that costs points | 7.1/10 | | Privacy policy | read, states 8 of the 8 things a reader expects, and has 1 clause that costs points | 8/10 | | Status page | www.devinstatus.com | 10/10 | | Changelog | published | 10/10 | | security.txt | not found | 0/10 | | People named | 3 named with their role, none with a public profile | 7/10 | | Support route | help pages | 5/5 | Not looked up: RDAP isn't requested under the 1 November 2026 rules. The product runs on devin.ai, registered 2022-12-06 per the devin listing's RDAP check of 8 October 2026. The launch post is on cognition.com, signed by the Cognition Team, and the Devin docs, pricing page and status page link the same company. The platform terms (30 June 2026) name Cognition AI, Inc. The website terms of use (13 May 2024) and platform terms were read for automated-access limits and have none. cognition.com's robots.txt disallows only /downloads/, devin.ai's only /top-secret and /assets/downloads/, docs.devin.ai's /cdn-cgi/ and /_next/ (with Content-Signal ai-input=yes), and www.devinstatus.com's /api/ and /embed/. security.txt is none per the devin listing's 8 October check, not re-read. The docs give a security address for vulnerability reports. No SWE-2 Flash entry in the Devin release notes, whose newest is 2 November 2026. ### Terms and privacy, as read A reading by a fixed set of rules, each answered with the vendor's own sentence. Not legal advice. **Terms of service** (https://cognition.com/legal/platform-terms-of-service), read 2026-10-08, dated 2026-06-30, states 6 of the 7 things a reader expects. - To know. Says it may use customer content to train or improve models, and gives an opt-out. "Cognition may use Customer Data for model training purposes and to improve and enhance the Services. If you subscribe to a paid Service Tier, you may opt out of this use (“Opt-Out”)." - To know. Restricts benchmarking or competitive use (costs points). "use the Services to create or develop any competing products or services, including to train competing artificial intelligence models except as expressly approved by Cognition in writing" - To know. Says access can be ended without notice or for any reason. "In addition, if you have a subscription, we may terminate the subscription at any time for any other reason." - To know. Requires arbitration or waives class actions. "11.1 IN THE EVENT A DISPUTE, CONTROVERSY, OR CLAIM ARISES OUT OF OR RELATING TO THESE TERMS (“DISPUTE”), THE DISPUTE WILL BE RESOLVED BY BINDING ARBITRATION RATHER THAN IN COURT." - Gives the date it was last updated. Last updated 2026-06-30. - Names the governing law or courts. The law of the State of New York. - States a limit on its liability. Capped at the greater of US$100 and the fees paid in the 6 months before the claim. - Says how changes to the terms are announced. Gives 30 days of notice before a change. - Not found in the text. Refers to a service level or uptime commitment. - Also in the text (2026-10-08). Either party's total liability is capped at the greater of the amounts paid in the six months before the event or 100 US dollars. "EXCEED THE GREATER OF (A) THE TOTAL AMOUNTS PAID TO COGNITION UNDER THIS AGREEMENT IN THE SIX MONTH PERIOD PRECEDING THE EVENT GIVING RISE TO THE CLAIM, OR (B) ONE HUNDRED DOLLARS (US$100)." - Also in the text (2026-10-08). Cognition may delete Customer Data when the terms or the subscription end, and the customer is responsible for its own backups. "Upon termination of these Terms or your subscription, we may at our option delete any Customer Data or other data associated with your account." - Also in the text (2026-10-08). The customer agrees not to process medical information or sensitive personal data, such as birth dates, bank account numbers and card numbers, through the Services. "You agree not to process any medical information or sensitive personal data such as social security numbers, birth dates, passport information, bank account, and credit card numbers in using the Services." **Privacy policy** (https://cognition.com/legal/privacy-policy), read 2026-10-08, dated 2026-03-09, states 8 of the 8 things a reader expects. - To know. Says it may use customer content to train or improve models, and no opt-out was found (costs points). "To customize your experience with our Services and otherwise improve our Services including, depending on the terms that apply to your use of the Services, using User Content to train, fine tune and improve the models that power our Services." - Gives the date it was last updated. Last updated 2026-03-09. - Says whether personal data is sold or shared for advertising. Says it does not sell personal data. - Gives a privacy contact. privacy@cognition.ai. - Says where data is transferred or stored. Relies on standard contractual clauses. - Also in the text (2026-10-08). Administrators of an enterprise or business account may be able to access a member's User Content and control the member's account. "In addition, administrators of any enterprise or business account may be able to access certain information associated with your account, including your User Content, and be able to control your account and such information." - Also in the text (2026-10-08). The privacy policy says users must be at least 18 years old to access the Services. "As set out in our Terms, users must be at least 18 years old in order to access the Services." ## Humans behind the project (researched 2026-11-04 from public pages, not part of the grade) ### Key people - Scott Wu, Co-founder and CEO - Steven Hao, Co-founder - Walden Yan, Co-founder - Note: Reused from data/humans/devin.json (checked 29 October 2026), since SWE-2 Flash is Cognition AI, Inc.'s model and the company is the same. The launch post of 4 November 2026 is signed by the Cognition Team and names no one, so no person is added for the model. Not re-read for this listing: the people, ownership and investors are as the devin file has them. Cognition's about page (read 28 October 2026) says its founding team holds 10 IOI gold medals but names nobody, and its Series E post of 8 September 2026 is signed by the Cognition team. Wikipedia names Scott Wu, Steven Hao and Walden Yan as founders and gives them as CEO, CTO and CPO. TechCrunch (29 May 2026) calls Scott Wu Cognition's CEO. - Cognition AI, Inc.: LinkedIn https://www.linkedin.com/company/cognition-ai-labs X https://x.com/cognition ### Ownership and investors - Valuation: $48B as announced (the source doesn't say pre- or post-money), Series E, September 2026 (source https://cognition.com/blog/series-e) - Backed by: Founders Fund, Lux Capital, 8VC, General Catalyst, Andreessen Horowitz, Accel, Avenir, Neo, Elad Gil, Definition Capital, Swish VC, Bain Capital Ventures, Hanabi Capital, D1 Capital Partners, Alpha Wave, Positive Sum, Vitruvian Partners, Conversion Capital, 137 Ventures, Soma Capital, Omri Casspi, Ribbit Capital, Atreides Management, Layer Global, Benchmark, Bessemer Venture Partners, Kleiner Perkins, Greylock, Lightspeed Venture Partners, Altimeter Capital, BOND, Meritech Capital, Valor Equity Partners, T. Rowe Price, DST Global, HOF Capital, BoxGroup, Battery Ventures, Stripes, Journey, A* Capital, Alkeon Capital, Diffusion ### Getting support - Help: https://devin.ai/support - Sources: https://cognition.com/blog/swe-2-flash, https://devin.ai/, https://cognition.com/about, https://cognition.com/team-spotlights, https://cognition.com/blog/introducing-devin, https://en.wikipedia.org/wiki/Cognition_AI, https://en.wikipedia.org/wiki/Scott_Wu ## Live (updated 2026-11-05 01:14 UTC) - Vendor status page: none, All Systems Operational - Always current: https://www.anchorterminal.com/api/v1/live/swe-2-flash.json ## Probe metrics Not measured yet. Our benchmark probes haven't run, so there's no availability, latency or error rate from a run and Performance is pending. Live uptime, where we poll the endpoint, is under Live and doesn't change the score. ## Strengths - Free in Devin Desktop and Devin CLI through 30 November 2026, and selectable per run with `devin --model` or `/model` - Cognition reports large gains over its base model, Terminal-Bench 4.0 from 0.9 to 20.0 and FrontierCode 2.0 from 3.2 to 24.0, with the full benchmark table in the post - Runs inside Devin CLI's sandbox, permissions and hooks, and enterprise admins can allow or block it per team - Cognition holds SOC 2 Type II and ISO/IEC 27001:2022 and publishes an SLA for contracts that include one - Status page with a Desktop Agent component, and in the 90 days to 4 November no incident names Desktop or the CLI, only a scheduled extension marketplace maintenance on 7 August ## Weaknesses - No model API, OpenAPI entry or model card. The Devin API's session agents list SWE-2 but not SWE-2 Flash - No per-token price, context length, output limit or reasoning settings published for it, and the docs' models pages don't list it yet - The weights are withheld for now, with a partner and researcher programme promised but not yet described - Customer data may be used for model training by default on self-serve plans, with the opt-out on paid plans only - The CLI's login token is persistent and doesn't expire by default, and no rate limit figures or retry guidance are published ## Before you call it (notes for agents) 1. There is no endpoint to call. Run Devin CLI single-turn with `devin -p` and pick the model with `--model`; the SWE-2 Flash model ID isn't documented yet, so check `/model` in the CLI first 2. Don't rely on the short name `swe`. The docs say it resolves to the latest model in the family, which may be SWE-2 rather than SWE-2 Flash 3. Plan for billing to start on 1 December 2026. The free period covers Devin Desktop and CLI only, not Devin cloud sessions 4. Treat Cognition's benchmark figures as the vendor's. No independent result for SWE-2 Flash was found 5. Keep `credentials.toml` private. The token in it doesn't expire by default and carries the signed-in user's access ## Connect Install: ```bash Install Devin CLI or Devin Desktop from https://devin.ai/download, sign in with `devin auth login`, then choose SWE-2 Flash with `/model` ``` ## Similar tools Ranked by shared capabilities, then score. Same-category tools with no shared capability key are listed last. | Tool | Grade | Score | Rank | Shared capabilities | x402 | Markdown | | --- | --- | --- | --- | --- | --- | --- | | OpenAI API | A | 83.9 | 4 | inference.llm | no | https://www.anchorterminal.com/tools/openai-api.md | | BlockRun.AI | A | 78.9 | 14 | inference.llm | yes | https://www.anchorterminal.com/tools/blockrun-ai.md | | Claude API | BB | 77.3 | 28 | inference.llm | no | https://www.anchorterminal.com/tools/anthropic-api.md | | GroqCloud | BB | 74.5 | 71 | inference.llm | no | https://www.anchorterminal.com/tools/groq.md | | Fireworks AI Inference | B | 69.3 | 228 | inference.llm | no | https://www.anchorterminal.com/tools/fireworks-inference.md | | Cloudflare Workers AI | B | 69.2 | 233 | inference.llm | no | https://www.anchorterminal.com/tools/cloudflare-workers-ai.md | ## Panel reviews (0) Reviewed by the Anchor panel (https://www.anchorterminal.com/reviewers/index.md): . Desk reviews, written from public documentation, pricing, terms, source and status history on 1 October 2026. No calls made. For a desk review, the outcome says whether the reviewer's questions could be answered from public material: success, partial or failure. How reviews work: https://www.anchorterminal.com/reviews/how-it-works.md ## Notable - Announced and made available in Devin Desktop and CLI on 4 November 2026, free through 30 November, at what Cognition says is one-fifth the price and three times the speed of SWE-2 (source: ) - Built on Thinking Machines' open-weight Inkling-Small (276B total, 12B active), with a mid-training stage of a few trillion tokens before reinforcement learning, Cognition's first (source: ) - Cognition is not sharing the weights publicly, citing cyber risk after raising ExploitBench from 11.3 to 34.8 per cent, and says a programme to share them with trusted partners and research labs will follow 'in a few weeks' (source: ) - Cognition's table gives Terminal-Bench 4.0 20.0, DeepSWE 1.1 60.1, FrontierCode 2.0 24.0, FrontierCode 1.1 Main 45.5, Terminal-Bench 2.1 89.2, SWE-Atlas QA 50.0 and SWE-Bench Pro v2 Hard 89.9. Vendor-reported, not our measurement (source: ) - The launch post says it outperforms all American open-weight models on leading evals, while its own table puts Meta's Muse Spark Open ahead on SWE-Atlas QA, 54.0 to 50.0 (source: ) - Not available through the Devin API: the v3 OpenAPI file lists SWE-2 Medium, High and Max as session agents and nothing for SWE-2 Flash (source: ) - #33 of 34 in Best model APIs and inference for AI agents: https://www.anchorterminal.com/best/inference/index.md - All 562 models comparisons: https://www.anchorterminal.com/compare/inference/index.md ## Compare - [Claude API vs SWE-2 Flash](https://www.anchorterminal.com/compare/anthropic-api-vs-swe-2-flash.md): BB 77.3 vs F 36.4 - [Antseed vs SWE-2 Flash](https://www.anchorterminal.com/compare/antseed-vs-swe-2-flash.md): C 54.1 vs F 36.4 - [BlockRun.AI vs SWE-2 Flash](https://www.anchorterminal.com/compare/blockrun-ai-vs-swe-2-flash.md): A 78.9 vs F 36.4 - [Cerebras Inference vs SWE-2 Flash](https://www.anchorterminal.com/compare/cerebras-inference-vs-swe-2-flash.md): B 68.6 vs F 36.4 - [Cloudflare AI Gateway vs SWE-2 Flash](https://www.anchorterminal.com/compare/cloudflare-ai-gateway-vs-swe-2-flash.md): B 67.8 vs F 36.4 - [Cloudflare Workers AI vs SWE-2 Flash](https://www.anchorterminal.com/compare/cloudflare-workers-ai-vs-swe-2-flash.md): B 69.2 vs F 36.4 - [Cohere Chat API (Command models) vs SWE-2 Flash](https://www.anchorterminal.com/compare/cohere-chat-vs-swe-2-flash.md): B 65.6 vs F 36.4 - [DeepInfra vs SWE-2 Flash](https://www.anchorterminal.com/compare/deepinfra-vs-swe-2-flash.md): B 63.7 vs F 36.4 - [DeepSeek API vs SWE-2 Flash](https://www.anchorterminal.com/compare/deepseek-api-vs-swe-2-flash.md): E 45.8 vs F 36.4 - [Glimpse vs SWE-2 Flash](https://www.anchorterminal.com/compare/elorian-glimpse-vs-swe-2-flash.md): D 48.8 vs F 36.4 - [Fireworks AI Inference vs SWE-2 Flash](https://www.anchorterminal.com/compare/fireworks-inference-vs-swe-2-flash.md): B 69.3 vs F 36.4 - [Gemini Developer API vs SWE-2 Flash](https://www.anchorterminal.com/compare/gemini-api-vs-swe-2-flash.md): B 62.2 vs F 36.4 - [GroqCloud vs SWE-2 Flash](https://www.anchorterminal.com/compare/groq-vs-swe-2-flash.md): BB 74.5 vs F 36.4 - [Jais 3 vs SWE-2 Flash](https://www.anchorterminal.com/compare/jais-3-vs-swe-2-flash.md): D 52.1 vs F 36.4 - [Meta Model API vs SWE-2 Flash](https://www.anchorterminal.com/compare/meta-model-api-vs-swe-2-flash.md): D 52.3 vs F 36.4 - [MiniMax API vs SWE-2 Flash](https://www.anchorterminal.com/compare/minimax-api-vs-swe-2-flash.md): C 55.3 vs F 36.4 - [Mistral AI API vs SWE-2 Flash](https://www.anchorterminal.com/compare/mistral-api-vs-swe-2-flash.md): B 65.7 vs F 36.4 - [Muse Spark Open vs SWE-2 Flash](https://www.anchorterminal.com/compare/muse-spark-open-vs-swe-2-flash.md): D 53.4 vs F 36.4 - [Nemotron 3.5 Super VL vs SWE-2 Flash](https://www.anchorterminal.com/compare/nemotron-3-5-super-vl-vs-swe-2-flash.md): C 59.3 vs F 36.4 - [Novita AI vs SWE-2 Flash](https://www.anchorterminal.com/compare/novita-ai-vs-swe-2-flash.md): D 53.9 vs F 36.4 - [OpenAI API vs SWE-2 Flash](https://www.anchorterminal.com/compare/openai-api-vs-swe-2-flash.md): A 83.9 vs F 36.4 - [OpenRouter vs SWE-2 Flash](https://www.anchorterminal.com/compare/openrouter-vs-swe-2-flash.md): B 66.6 vs F 36.4 - [Prism Inference vs SWE-2 Flash](https://www.anchorterminal.com/compare/prism-inference-vs-swe-2-flash.md): C 59.6 vs F 36.4 - [Router.com (Ramp Router) vs SWE-2 Flash](https://www.anchorterminal.com/compare/ramp-router-vs-swe-2-flash.md): D 53.7 vs F 36.4 - [Beam 1.0 vs SWE-2 Flash](https://www.anchorterminal.com/compare/reflection-beam-vs-swe-2-flash.md): D 53.4 vs F 36.4 - [RouterPlus vs SWE-2 Flash](https://www.anchorterminal.com/compare/routerplus-vs-swe-2-flash.md): D 50.3 vs F 36.4 - [SambaCloud vs SWE-2 Flash](https://www.anchorterminal.com/compare/sambanova-vs-swe-2-flash.md): B 64.9 vs F 36.4 - [SiliconFlow vs SWE-2 Flash](https://www.anchorterminal.com/compare/siliconflow-vs-swe-2-flash.md): D 46.8 vs F 36.4 - [Simplismart vs SWE-2 Flash](https://www.anchorterminal.com/compare/simplismart-vs-swe-2-flash.md): F 29.5 vs F 36.4 - [StepFun API vs SWE-2 Flash](https://www.anchorterminal.com/compare/stepfun-api-vs-swe-2-flash.md): E 44.3 vs F 36.4 - [SWE-2 Flash vs Tencent Cloud TokenHub](https://www.anchorterminal.com/compare/swe-2-flash-vs-tencent-tokenhub.md): F 36.4 vs C 59.5 - [SWE-2 Flash vs Together AI Inference](https://www.anchorterminal.com/compare/swe-2-flash-vs-together-inference.md): F 36.4 vs B 67.8 - [SWE-2 Flash vs Vercel AI Gateway](https://www.anchorterminal.com/compare/swe-2-flash-vs-vercel-ai-gateway.md): F 36.4 vs B 67.8 ## Verify this listing For the vendor. The badge or a plain link to this page verifies the listing, from a page on cognition.com or one of its subdomains. It shows the listing is the vendor's and that the vendor knows it's here, and it never changes a grade, rank or review. The vendor sends the page's address to `POST https://www.anchorterminal.com/api/v1/verify` as `{"slug": "swe-2-flash", "url": "…"}`, or calls the `verify_listing` tool at https://www.anchorterminal.com/mcp. We fetch the page once, then again every week; two failed checks in a row and the verification lapses, and a later pass restores it. What we check: https://www.anchorterminal.com/builders/index.md#verify HTML badge: ```html SWE-2 Flash on Anchor Terminal ``` Markdown badge, for a README: ```markdown [![SWE-2 Flash on Anchor Terminal](https://www.anchorterminal.com/badges/swe-2-flash.svg)](https://www.anchorterminal.com/tools/swe-2-flash) ``` Plain link: ```html SWE-2 Flash on Anchor Terminal ``` ## Share this listing For the vendor. Sharing assets for social media, two PNGs of 1200 × 630 that say SWE-2 Flash is listed on Anchor Terminal, with the vendor's logo and this page's address and no grade or score. - Dark: https://www.anchorterminal.com/assets/share/swe-2-flash-dark.png - Light: https://www.anchorterminal.com/assets/share/swe-2-flash-light.png --- Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp