Open-weight model data · sweep
Which open-weight model should I use?
Provider prices, spreads, and throughput for identical open weights — measured and dated, never marketed.
- DeepSeek V3.x
- DeepSeek R1
- Qwen3 235B
- Qwen3 Coder
- Llama 4 Maverick
- Kimi K2
- GLM 4.7
- GPT-OSS 120B
- Gemma 3 27B
- MiniMax M3
Head-to-heads: DeepSeek V3.x vs Qwen3 · DeepSeek R1 vs Kimi K2 · Qwen3 Coder vs GLM
Three tools, zero opinions for sale
Price Tracker
Verified $/M-token prices across providers for one model, side by side, spread vs cheapest highlighted.
Cost Calculator
Your in/out/cache/batch mix becomes a true monthly bill per provider.
Self-Host TCO
GPU rental plus ops vs API break-even at your traffic level.
Latest price moves
No price moves recorded yet — the feed fills after two sweep cycles diff against a baseline.
Open Weight Weekly
Price wars, new checkpoints, spread alerts. One email each Friday, numbers only.
Data sections carry no sponsorships. Unsubscribe anytime.