Qwen3.8 27B
Alibaba's Qwen3.8 27B (Apache 2.0) is a dense multimodal model with thinking and instruct modes in one checkpoint, aimed at agentic coding. Free API on Groq as qwen/qwen3.8-27b (131K context there, 30 RPM, 1,000 requests/day). Locally, ollama run qwen3.8:27b pulls an 18 GB Q4_K_M build — set num_ctx yourself (for example 32768), because the native 262K context needs far more memory for the KV cache.
Type: local, api
Context: 262,144 tokens
Use cases: coding, reasoning, vision, general
Free API
| Provider | Free | RPM | Daily | Checked |
|---|---|---|---|---|
| groq | yes | 30 | 1000 | 2026-10-10 |
Limits checked against official docs on 2026-10-10. Rates change without notice. Not affiliated with providers unless noted. Verify before production use.