AFM 0.9.19
Expanded Qwen, GLM, and DeepSeek support; native checkpoint conversion; OpenAI Responses, Anthropic Messages, and image APIs; stronger cache and runtime reliability; an updated WebUI; and detailed cross-machine qualification evidence.
Models + performance
Improves Qwen 3.8 Flash Next, GLM 5.3 Flash, and DeepSeek V4 Flash 0731 with MTP, execution optimizations, sparse verification, and bounded-memory fixes.
Native checkpoint conversion
Adds Qwen Next conversion controls, mapped or resident n-gram storage, and safer storage preflight and resume handling.
More API compatibility
Adds OpenAI Responses, Anthropic Messages and token counting, text completions, and image generation and editing for supported backends.
Cache + runtime reliability
Improves prefix-cache validation, batch dispatch, capability reporting, assistant prefill, tool-call collection, rate limiting, and fatal startup errors.
Updated WebUI + packaging
Updates the bundled WebUI, keeps the runtime-control side panel disabled pending redesign, and strengthens resource and Foundation Models build checks.
Evaluation + observability
Adds feature-specific prompts, complete AI-judge reporting, clearer result categories, live TTFT measurement, and more precise failure attribution.
Read the release notes on GitHub ↗Open the release test evidence →
