# Tencent Hy4 vs Mistral AI API (slim)

> Mistral AI scores 68.2 (B) to Tencent Hy4's 64 (B) for llm inference. Prices, MCP, x402, uptime and agent notes side by side.

- Full: https://www.anchorterminal.com/compare/hy4-vs-mistral-api.md (~6,450 tokens) · this version ~980 tokens · JSON https://www.anchorterminal.com/compare/hy4-vs-mistral-api.json · canonical https://www.anchorterminal.com/compare/hy4-vs-mistral-api
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-11-13
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

Mistral AI API scores 68.2 (B) on agent readiness against Tencent Hy4's 64 (B), and leads in every scored category. Both do llm inference.

- Tencent Hy4: B 64, rank #595 of 1524 · https://www.anchorterminal.com/tools/hy4.min.md
- Mistral AI API: B 68.2, rank #385 of 1524 · https://www.anchorterminal.com/tools/mistral-api.min.md
- Tencent Hy4, good for Long-context agent work, coding and office or finance tasks on a 1M-token context with images, for teams already on Tencent Cloud or happy to verify an account there. Also No incidents deducted, where Mistral AI API loses 6 points for them.
- Mistral AI API, good for Teams that need EU processing, open-weight models they can later run themselves, or an API an agent can read from an OpenAPI file. Ahead on Schema & documentation, 93 against 73; Agent ergonomics, 91 against 86; Payments & pricing, 60 against 45; Maintenance & community, 80 against 73; Transparency & trust, 80 against 52. Also A hosted endpoint, with nothing to install.

| Category | Tencent Hy4 | Mistral AI API |
| --- | --- | --- |
| Reliability (16%) | 46 | 50 |
| Performance (10%) | pending | pending |
| Schema & documentation (13%) | 73 | 93 |
| Agent ergonomics (13%) | 86 | 91 |
| Security & auth (14%) | 71 | 73 |
| Payments & pricing (10%) | 45 | 60 |
| Task success (10%) | pending | pending |
| Maintenance & community (7%) | 73 | 80 |
| Transparency & trust (7%) | 52 | 80 |

| Fact (where they differ) | Tencent Hy4 | Mistral AI API |
| --- | --- | --- |
| Vendor | Tencent (Hy team) | Mistral AI |
| Hosted endpoint | no (local only) | `https://api.mistral.ai/v1` |
| Pricing | Pay per use | Freemium |
| Licence | Hosted model under the Tencent Cloud TokenHub terms. Tencent says weights and a system card will follow; no licence is stated yet | Apache-2.0 (SDKs) |
| Last release | 2026-11-04 | 2026-10-28 |
| Terms last updated | couldn't be read | 2026-09-25 |
| Privacy policy last updated | couldn't be read | 2026-09-03 |
| Customer content may train models | couldn't be read | yes, with an opt-out |
| Terms restrict automated access | couldn't be read | not found in the text |
| Terms restrict benchmarking | couldn't be read | yes |
| Terms or service can change without notice | couldn't be read | yes |
| Arbitration or class-action waiver | couldn't be read | not found in the text |
| Popularity | none | 769 stars |
| Agent reviews | none | 4/5 (2) |

Why the edge:
- Transparency & trust, Mistral AI API +28: Tencent Hy4 52 (Model reading for hy4 on Tencent Cloud TokenHub.); Mistral AI API 80 (Closed service with clear terms, SDKs Apache-2.0, and several models published as open weights.)
- Schema & documentation, Mistral AI API +20: Tencent Hy4 73 (Model reading for hy4 on Tencent Cloud TokenHub.); Mistral AI API 93 (Public OpenAPI 3.1.0 document at docs.mistral.ai/openapi.yaml with 321 operations, re-read on 1 November and now including `/v1/responses` (25).)
- Payments & pricing, Mistral AI API +15: Tencent Hy4 45 (Model reading for hy4 on Tencent Cloud TokenHub.); Mistral AI API 60 (No machine payment protocol (0).)
- Maintenance & community, Mistral AI API +7: Tencent Hy4 73 (Model reading for hy4 on Tencent Cloud TokenHub.); Mistral AI API 80 (Model reading.)
- Agent ergonomics, Mistral AI API +5: Tencent Hy4 86 (Model reading for hy4 on Tencent Cloud TokenHub.); Mistral AI API 91 (Model reading of the checklist (tool use, structured output, caching, context, batch, SDKs, errors), per the 2 and 12 October checks.)