v0.13.2
CaviraOSS/LongMemoryv0.13.2Sep 29, 2026by tashfeenahmed
Upgrading?
AI Summary
This release fixes compatibility for Claude Code and other agents with free models that have lower output token limits by allowing the router to learn and respect those limits from refusal responses.
Key Highlights
- Models with smaller output limits (e.g., Groq, Ollama) now work correctly with Claude Code.
- The router learns each model's output limit from the first refusal and retries within that limit.
- Fixes apply across all endpoints: Claude/Anthropic, OpenAI chat, Responses, and Gemini/Ollama.
New Features
- Automatic learning of model output limits from refusal responses
- Retry logic that respects dynamic output token caps
Full Release Notes
FreeLLMAPI v0.13.2 keeps Claude Code working on models that cap their output below what Claude Code asks for. ## Fixes - **Models with a smaller output limit keep working with Claude Code.** Claude Code asks for up to 128,000 output tokens. Some free models stop at 65,536 and turned the request down: Groq's gpt-oss-20b and Ollama Cloud's Nemotron 3 Super and Ultra. The router used to set those models aside and wrongly marked them as unable to use tools. Now it learns each model's output limit from the first refusal, retries straight away within that limit, and keeps within it from then on. This covers every endpoint: Claude/Anthropic, OpenAI chat, Responses and Gemini/Ollama. (#1364) ## Updating Download the installer for your platform below, or use `ghcr.io/tashfeenahmed/freellmapi:v0.13.2` for Docker. The `latest` Docker tag also points to this release. Desktop users on v0.13.1 will see this update in the Updates panel. Existing keys and profiles are retained. This release has no database migrations. [Full changelog](https://github.com/tashfeenahmed/freellmapi/compare/v0.13.1...v0.13.2) --- ā Like the free router? [Go Premium](https://freellmapi.co/?utm_source=github&utm_medium=release&utm_campaign=premium&utm_content=v0.13.2#pricing), the live signed catalog, $19/yr, cancel anytime.
Related Repositories
openclaw/openclawā
392,696
The AI that really does things. Any OS. Any Platform. The lobster way. š¦
nilbuild/developer-roadmapā
370,067
Interactive roadmaps, guides and other educational content to help developers grow in their careers.
morluto/reaā
357,700
Reverse engineer anything with agents, from app behavior down to native binaries.
anomalyco/opencodeā
219,670
The open source coding agent.
n8n-io/n8nā
208,251
Fair-code workflow automation platform with native AI capabilities. Combine visual building with custom code, self-host or cloud, 400+ integrations.