mlc-ai/web-llm v0.2.87

on GitHub

What's Changed

• Upgrade @mlc-ai/web-runtime to 0.28.0-dev1
• Update model binaries to v0_2_87, including subgroup variants
• Improve Gemma 4 q4f32 prefill performance
• Fix sliding-window KV cache handling
• Fix JSON schema generation for Qwen3.5
• Fix Gemma 3 context handling
• Update examples and documentation to 0.2.87

Full Changelog: v0.2.86...v0.2.87