Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Get help setting up a custom Dedicated Endpoints.
Talk with our engineer to get a quote for reserved GPU instances with discounts.
README
License: apache-2.0Evaluation (pass@1, 40 held-out test tasks)
| checkpoint | pass@1 |
|---|---|
| base (step 0) | 0.250 |
| this release (step 200) | 0.350 |
| run peak (step 175, not checkpointed) | 0.375 |
Caveat: eval used a gpt-5.4-nano user-simulator and NL-assertion judge, which is
weaker than the standard gpt-4.1 user-sim. These numbers are not comparable to
published τ²-bench results — they are only meaningful relative to the same-sim baseline
(~0.25) reported here.
Usage
Standard transformers / vLLM generation. Tool calls use the Hermes format
(--tool-call-parser hermes in vLLM). Use the model's default chat template.
Model provider
reinforcelabs
Model tree
Base
Qwen/Qwen3-4B-Instruct-2507
Fine-tuned
this model
Modalities
Input
Text
Output
Text
Pricing
Dedicated Endpoints
View detailsSupported Functionality
Model APIs
Dedicated Endpoints
Container
More information