Precision bait
Distinct sellers, not rows. A cheaper lower-quant endpoint is a trade, not a saving. See /spreads/.
As of 2026-09-19. Public lens does not apply; these are price-shape facts.
| Model | What we measured |
|---|---|
| DeepSeek V4.1 Flash | serving precision differs across offers — cheapest may be worse quant |
| GLM 5.3 Flash | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V4 Flash Vision Exp | serving precision differs across offers — cheapest may be worse quant |
| GLM 5.3 | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.8 27B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.8 2.4T A95B | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V4 Pro 0813 | serving precision differs across offers — cheapest may be worse quant |
| Nemotron 3.5 Lightning | serving precision differs across offers — cheapest may be worse quant |
| Muse Glimmer 30B | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V4 Flash 0731 | serving precision differs across offers — cheapest may be worse quant |
| Inkling Small | serving precision differs across offers — cheapest may be worse quant |
| Ling 3.0 Flash | serving precision differs across offers — cheapest may be worse quant |
| Inkling | serving precision differs across offers — cheapest may be worse quant |
| Kimi K3 | serving precision differs across offers — cheapest may be worse quant |
| Hy3 | serving precision differs across offers — cheapest may be worse quant |
| GLM 5.2 | serving precision differs across offers — cheapest may be worse quant |
| Kimi K2.7 Code | serving precision differs across offers — cheapest may be worse quant |
| Nemotron 3 Ultra | serving precision differs across offers — cheapest may be worse quant |
| MiniMax M3 | serving precision differs across offers — cheapest may be worse quant |
| Step 3.7 Flash | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.6 35B A3B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.6 27B | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V4 Pro 0423 | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V4 Flash 0423 | serving precision differs across offers — cheapest may be worse quant |
| MiMo-V2.5-Pro | serving precision differs across offers — cheapest may be worse quant |
| MiMo-V2.5 | serving precision differs across offers — cheapest may be worse quant |
| Kimi K2.6 | serving precision differs across offers — cheapest may be worse quant |
| GLM 5.1 | serving precision differs across offers — cheapest may be worse quant |
| Gemma 4 26B A4B | serving precision differs across offers — cheapest may be worse quant |
| Gemma 4 31B | serving precision differs across offers — cheapest may be worse quant |
| MiniMax M2.7 | serving precision differs across offers — cheapest may be worse quant |
| Nemotron 3 Super | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.5-9B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.5-35B-A3B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.5-27B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.5-122B-A10B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3.5 397B A17B | serving precision differs across offers — cheapest may be worse quant |
| MiniMax M2.5 | serving precision differs across offers — cheapest may be worse quant |
| GLM 5 | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 Coder Next | serving precision differs across offers — cheapest may be worse quant |
| Kimi K2.5 | serving precision differs across offers — cheapest may be worse quant |
| GLM 4.7 Flash | serving precision differs across offers — cheapest may be worse quant |
| GLM 4.7 | serving precision differs across offers — cheapest may be worse quant |
| Nemotron 3 Nano 30B A3B | serving precision differs across offers — cheapest may be worse quant |
| GLM 4.6V | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V3.2 | serving precision differs across offers — cheapest may be worse quant |
| Kimi K2 Thinking | serving precision differs across offers — cheapest may be worse quant |
| MiniMax M2 | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 VL 8B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 VL 30B A3B Thinking | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 VL 30B A3B Instruct | serving precision differs across offers — cheapest may be worse quant |
| GLM 4.6 | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 VL 235B A22B Thinking | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 VL 235B A22B Instruct | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V3.1 Terminus | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 Next 80B A3B Instruct | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V3.1 | serving precision differs across offers — cheapest may be worse quant |
| gpt-oss-120b | serving precision differs across offers — cheapest may be worse quant |
| gpt-oss-20b | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 Coder 30B A3B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 30B A3B Instruct 2507 | serving precision differs across offers — cheapest may be worse quant |
| GLM 4.5 Air | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 235B A22B Thinking 2507 | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 Coder 480B A35B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 235B A22B Instruct 2507 | serving precision differs across offers — cheapest may be worse quant |
| Mistral Small 3.2 24B | serving precision differs across offers — cheapest may be worse quant |
| MiniMax M1 | serving precision differs across offers — cheapest may be worse quant |
| R1 0528 | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 30B A3B | serving precision differs across offers — cheapest may be worse quant |
| Qwen3 14B | serving precision differs across offers — cheapest may be worse quant |
| Llama 4 Maverick | serving precision differs across offers — cheapest may be worse quant |
| Llama 4 Scout | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V3 0324 | serving precision differs across offers — cheapest may be worse quant |
| Gemma 3 27B | serving precision differs across offers — cheapest may be worse quant |
| DeepSeek V3 | serving precision differs across offers — cheapest may be worse quant |
| Llama 3.3 70B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Llama 3.2 3B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Qwen2.5 72B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Llama 3 8B Lunaris | serving precision differs across offers — cheapest may be worse quant |
| Llama 3.1 70B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Llama 3.1 8B Instruct | serving precision differs across offers — cheapest may be worse quant |
| Mistral Nemo | serving precision differs across offers — cheapest may be worse quant |
| ReMM SLERP 13B | serving precision differs across offers — cheapest may be worse quant |
| MythoMax 13B | serving precision differs across offers — cheapest may be worse quant |