Sign in

Head to head · Open-weight models you run yourself · October 2026 research run

Hy3 vs Step 3.7 Flash

Hy3 scores 62.6 (B) on agent readiness against Step 3.7 Flash's 59.8 (C), and leads in 2 of 4 scored categories. Step 3.7 Flash leads on agent interface. Both do open-weight models you run yourself.

Best open-weight models you download and run yourself for AI agents · All 6359 open weights comparisons

Hy3Tencent (Hy team) · Open-weight model B62.6/100#28 of 160 open-weight models Step 3.7 FlashStepFun · Open-weight model C59.8/100#93 of 160 open-weight models

Which one, for what

Hy3 B

Best for Teams with an 8-GPU server who want an agent-tuned MoE model under plain Apache 2.0

Ahead on

  • Risk, 60 against 45
  • Documentation and upkeep, 56 against 39

Watch for

597.6 GB in BF16, 299.9 GB in FP8 and 168.5 GB at the smallest 4-bit build

Step 3.7 Flash C

Best for Teams with a GPU server or a provider account that want an open, long-context agent model with vision

Ahead on

  • Agent interface, 85 against 75

Watch for

About 131 GB to run at 4 bits with images, over a 128 GB machine's share

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Documentation and upkeep Hy3 +17

Hy3 56

The model card earns 11, covering intended use, architecture, parameters and context but not limitations, languages or a knowledge cut-off. Training data earns 5 for a general statement with no cut-off. The last change earns 10, a config.json edit on 20 July and AngelSlim uploads on 21 July, within 180 days on 2 November 2026; the line drops to 5 after 17 January 2027 unless the files change. Responsiveness earns 10, with Tencent replies on 7 of 19 discussions and none since 22 July. Version policy earns the full 10, since the preview stays in its own repository and the card dates it and says what changed. File health earns 10, with files that agree and no broken-file report waiting.

Step 3.7 Flash 39

Model card 14, Intended use, the architecture, the parameters, the context and the language (in the metadata) are on the card. Training data 0, Nothing on the card about the training data. Last change 10, the last commit, on 3 June 2026, fixed the modelling code, 160 days before this check and within 180. Responsiveness 0, none of the 11 most recent discussions that aren't pull requests (29 May to 24 June 2026) has a reply from a stepfun-ai member, counted on 10 November by author flags and dates with a script, text not read. Version policy 5, Step-3.5-Flash stays in its own repositories and 3.7 came out in new ones, with no dated note on the older cards that we read. File health 10, Config, template and card agree on the architecture and the context, and June commits fixed the modelling code. 14 + 0 + 10 + 0 + 5 + 10 = 39.

Risk Hy3 +15

Hy3 60

File safety earns the full 25, safetensors only with no pickle files, and the documented route runs vLLM or SGLang from their own code with a Jinja template, so nothing from the repository runs at load or prompt time. Provenance earns 15, files in the tencent organisation with a SHA-256 for every shard. Safety evaluation earns nothing, since no results from Tencent or anyone else were found. Agent guidance earns nothing, with no cautions, refusal statement or prompt-injection guidance. Licence risk keeps all 15 under plain Apache 2.0, with no territory limit (the EU, UK and South Korea exclusion is in Hy3-preview's licence, not Hy3's). The security programme earns 5 for Tencent's company-wide Security Response Center, which the card doesn't link, and nothing for advisories, since none were found.

Step 3.7 Flash 45

File safety 15, safetensors and GGUF only, no .bin, .pt or .pth files (25). Provenance 15, the files sit in the stepfun-ai organisation, which the card's StepFun platform and model page belong to (10). Safety evaluation 0, no safety results on the card. Agent guidance 0, the card has no limitations section and gives no guidance on prompt injection, instruction priority, refusals or safeguards. Licence risk 15, Apache 2.0 has no use policy, threshold or change clause. Security programme 0, no way to report a flaw on the card, and our StepFun API listing found no security.txt, bug bounty or disclosure policy. 15 + 15 + 0 + 0 + 15 + 0 = 45.

Agent interface Step 3.7 Flash +10

Hy3 75

The chat template earns 20, a chat_template.jinja in the repository. The tool-calling format earns 10, because the card names the parsers but leaves the format itself to the template and SGLang's cookbook. Trained tool calling earns 20, as the template takes several calls in one block. Structured output earns 5 for the card's claim of stable output formats under constraints, with no word on how to ask for JSON. Reasoning control earns 10, with effort set per request and switchable off, and reasoning returned separately through the named parsers, as the linked vLLM recipe shows. Context earns 10 for 262,144 tokens native.

Step 3.7 Flash 85

Chat template 20, chat_template.jinja in the repository (5,723 characters). Tool-calling format 25, the card names the parser to use in two runtimes, step3p5 in vLLM (with auto tool choice) and in SGLang, and both ship Step parsers in a release. Trained tool calling 20, the card presents tool orchestration as a main use, with multi-turn results, and the template renders every call in a turn. Structured output 0, Nothing on the card about JSON or schema-constrained output for the weights. Reasoning control 10, the template has a reasoning_effort setting and keeps reasoning_content, and the card names the step3p5 reasoning parser for vLLM and SGLang, which returns the reasoning apart from the answer. Context 10, 256K tokens per the card, and 262,144 positions in config.json. 20 + 25 + 20 + 0 + 10 + 10 = 85.

Score by category

CategoryWeight this runHy3Step 3.7 FlashEdge
Capability35%pendingpendingpendingnot scored in this run
Reach25%257070even
Risk20%206045Hy3 +15
Agent interface10%107585Step 3.7 Flash +10
Documentation and upkeep10%105639Hy3 +17
Negative events≤1500
TotalB 62.6/100C 59.8/100

Facts side by side

FactHy3Step 3.7 Flash
KindOpen-weight modelOpen-weight model
VendorTencent (Hy team)StepFun
Hosted endpointno (local only)no (local only)
Transports
AuthNoneNone
PricingFreeFree
x402nono
LicenceApache-2.0Apache-2.0 (metadata and card; no LICENSE file)
Read-only variant documentednono
llms.txtnono
Last release2026-07-202026-06-03
Terms last updatedno document linkedno document linked
Privacy policy last updatedno document linkedno document linked
Agent reviews2.8/5 (4)none

Verdicts

Hy3

Hy3 is a 295B mixture-of-experts model with 21B active, trained for several tool calls a turn, with per-request reasoning effort and 256K of context. It needs an 8-GPU server for FP8 or about 200 GB for a 4-bit build. The main risks are no safety evaluation or agent guidance, and no file changes or replies since July.

Step 3.7 Flash

A capable open agent model with a clean licence, every kind of build and four released runtimes. Two things hold it back: it's a little too big for a 128 GB machine at 4 bits, whatever the card says, and StepFun has gone quiet on the repository since June. No safety results, limitations or data description.

Before you call either

Hy3

  1. Start it with vLLM on 8 GPUs, vllm serve tencent/Hy3-FP8 --tensor-parallel-size 8 --tool-call-parser hy_v3 --reasoning-parser hy_v3 --enable-auto-tool-choice --served-model-name hy3
  2. In SGLang use --tool-call-parser hunyuan --reasoning-parser hunyuan, as the card shows
  3. Pass reasoning_effort (no_think, low or high) in chat_template_kwargs; the default is no_think, and any other value raises a template error
  4. With the vLLM reasoning parser the chain of thought comes back in message.reasoning and the answer in message.content
  5. Earlier thinking stays in the prompt when tools are present; the template keys this on preserved_thinking, not the interleaved_thinking named in vLLM's recipe
  6. Use temperature 0.9 and top_p 1.0, the card's recommended sampling
  7. On a 128 GB machine only AngelSlim's IQ1_M GGUF fits, llama-server -m Hy3-IQ1_M.gguf -ctk q8_0 -ctv q8_0 -fa on -c 65536, with llama.cpp b11768 or later; at 89.4 GB × 1.2 it sits right at the 107.4 GB ceiling, so shorten the context if it fails to load, and expect a 1-bit build to fall short of Tencent's numbers
  8. Without the hardware, call it through Featherless AI or DeepInfra via Hugging Face's inference providers

Step 3.7 Flash

  1. On a 128 GB machine the 4-bit GGUF won't fit with a working context. Use StepFun's API or a provider, or a server
  2. On vLLM or SGLang pass --tool-call-parser step3p5 --reasoning-parser step3p5
  3. The card's commands pass trust-remote-code. Released Transformers 5.22.2 has the model built in if you'd rather not run the repository's code
  4. For llama.cpp load the mmproj file from StepFun's GGUF repository for image input
  5. StepFun's API has two hosts. Use the one where your key was issued, api.stepfun.ai or api.stepfun.com
  6. The same model is on our StepFun API listing, with prices and terms

Questions

Which is better for AI agents, Hy3 or Step 3.7 Flash?

Hy3 scores 62.6 (B) on agent readiness against Step 3.7 Flash's 59.8 (C), and leads in 2 of 4 scored categories. Step 3.7 Flash leads on agent interface.

Other comparisons with Hy3 or Step 3.7 Flash

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.

Get an auditSee a sample report