minimax-m2.5 API — compare providers
The same model (minimax-m2.5) is served by 7 providers; Venice is the cheapest at $1.22/1M blended — the priciest (Minimax) costs 146% more. On quality, minimax-m2.5 scores 1391 on the LMArena (Chatbot Arena) — so the cheapest endpoint is also the best value per dollar here.
Quality and intelligence scores are attached to the model itself — identical whichever provider serves it. Verified 2026-09-04.
| Rank | Provider | Input /1M | Output /1M | Blended /1M | Value (Elo/$) |
|---|---|---|---|---|---|
| 1 | Venice | $0.27 | $0.95 | $1.22 ← cheapest | 1,140 ← best value |
| 2 | AtlasCloud | $0.295 | $1.2 | $1.495 | 930 |
| 3 | DigitalOcean | $0.3 | $1.2 | $1.5 | 927 |
| 4 | Friendli | $0.3 | $1.2 | $1.5 | 927 |
| 5 | SiliconFlow | $0.3 | $1.2 | $1.5 | 927 |
| 6 | Novita AI | $0.3 | $1.2 | $1.5 | 927 |
| 7 | Minimax | $0.6 | $2.4 | $3 | 464 |
Value = model Elo ÷ blended price per 1M tokens. Since the model's quality is identical across providers, the cheapest endpoint is also the best value per dollar.
Prices are per 1M tokens (USD). "Blended" = input + output, for coarse ranking. Same underlying model, 7 serving providers, 146% spread top to bottom. Context window: 204,800 tokens (32,768 max output) — see how it ranks in biggest context windows. See how this compares across the catalogue in same model, different price, or find the best quality per dollar in the value ranking.
Cost per 1,000 requests by workload
What each provider actually bills for a representative job, not just the sticker price.
| Provider | Chatbot 1000 in / 500 out | RAG / long context 8000 in / 500 out | Batch summarize 4000 in / 1000 out |
|---|---|---|---|
| Venice | $0.74 | $2.64 | $2.03 |
| AtlasCloud | $0.90 | $2.96 | $2.38 |
| DigitalOcean | $0.90 | $3.00 | $2.40 |
| Friendli | $0.90 | $3.00 | $2.40 |
| SiliconFlow | $0.90 | $3.00 | $2.40 |
| Novita AI | $0.90 | $3.00 | $2.40 |
| Minimax | $1.80 | $6.00 | $4.80 |
Estimate your own workload
Input tokens/request: Output tokens/request: Requests:
| Provider | Estimated cost (USD) |
|---|
Self-host / quantized builds
Open weights under MiniMax Model License — you can run minimax-m2.5 on your own hardware instead of paying per token. Quantized builds lower the memory footprint; whether it fits your GPU or Mac depends on the model's size. Formats: GGUF for llama.cpp / Ollama / LM Studio, AWQ and GPTQ for GPU serving, MLX for Apple silicon.
- GGUF builds llama.cpp · Ollama · LM Studio
- AWQ builds GPU serving (vLLM)
- GPTQ builds GPU serving
- MLX builds Apple silicon
- Ollama library one-command local run
Links search the live model hubs (Hugging Face, Ollama) so they track new builds as they appear — we don't host weights. Which formats exist depends on what the model owner and community have published.