MiniMax M3 vs Qwen3-VL 235B
Two open-weight chat models on the same endpoint and API key, so switching between them is a one-word change. Below: what each costs for three realistic workloads, how they differ, and measured speed.
Inference APIs prices are read live from the price list.
- Cost: Qwen3-VL 235B is cheaper in all three workloads, by 27–32%.
- Speed (measured): MiniMax M3 1.3 s to first token and 72 tok/s; Qwen3-VL 235B 1.28 s and 11 tok/s.
Side by side
| MiniMax M3 | Qwen3-VL 235B | |
|---|---|---|
| Served by | Inference APIs | Inference APIs |
| Model author | MiniMax | Alibaba Qwen |
| Model id | MiniMaxAI/MiniMax-M3 | Qwen/Qwen3-VL-235B-A22B-Instruct |
| Input price | $0.39 / 1M tokens | $0.26 / 1M tokens |
| Output price | $1.56 / 1M tokens | $1.15 / 1M tokens |
| Context window | 512K tokens | 262K tokens |
| Capabilities | Chat, Reasoning, Tool calling, JSON mode | Vision, OCR, Chat, Tool calling, JSON mode |
| Weights | Open | Open |
| Status | Available | Available |
| Limits | No per-minute or per-day request or token caps; usage draws on a prepaid balance | No per-minute or per-day request or token caps; usage draws on a prepaid balance |
| Time to first token | 1.3 s | 1.28 s |
| Output speed | 72 tokens / s | 10.7 tokens / s |
What three workloads cost
List prices applied to the same work. The cheaper side of each row is in bold.
| Workload | MiniMax M3 | Qwen3-VL 235B | Difference |
|---|---|---|---|
| Chat assistant — 10,000 turns of 800 tokens in, 300 out | $7.80 | $5.53 | 29% |
| Long prompts or RAG — 10,000 requests of 8,000 in, 500 out | $39.00 | $26.55 | 32% |
| Generation-heavy — 10,000 requests of 500 in, 2,000 out | $33.15 | $24.30 | 27% |
Qwen3-VL 235B is cheaper in all three workloads, by 27–32%. Reasoning models bill their thinking as output tokens, so real output counts run higher than the visible answer; treat the generation-heavy row as a floor for them.
When to choose which
- You need more than 262K tokens of context (it takes 512K)
- You stream long answers: 72 tokens per second against 11
- You want a large-context agent model priced near the Flash tier
- Text-only work; the text models here are cheaper
- You need to send images; MiniMax M3 is text-only
- Cost is the deciding factor; it is cheaper in every workload above
- Reading screenshots, invoices, receipts and scanned pages into text or JSON
- Agents that need to look at an image and then call a tool
Price, context and speed are measurable; answer quality on your task is not something a table can settle. Both take the same request, so the honest test is to run your own prompts through each in the playground.
Measured speed
Medians of three streamed runs on 2026-09-17, public endpoint, a prompt that asks for about 120 words. Time to first token is the first token of any kind, reasoning included; output speed counts every generated token from that point. Reasoning models then think before the visible answer starts: MiniMax M3 began answering after 2.74 s and Qwen3-VL 235B after 1.28 s at default settings, which you can shorten with the reasoning controls on each model page. Expect ±30% with time of day. Method and raw numbers: speed measurements.
Trying both
Same endpoint, same key. Change one string:
for MODEL in "MiniMaxAI/MiniMax-M3" "Qwen/Qwen3-VL-235B-A22B-Instruct"; do
curl -s https://api.inferenceapis.com/v1/chat/completions \
-H "Authorization: Bearer $INFERENCE_API_KEY" -H "Content-Type: application/json" \
-d "{\"model\": \"$MODEL\", \"max_tokens\": 500, \"messages\": [{\"role\": \"user\", \"content\": \"Summarise the plot of Hamlet in two sentences.\"}]}" \
| jq -r '.model, .choices[0].message.content, .usage'
doneFAQ
Which is cheaper, MiniMax M3 or Qwen3-VL 235B?
Qwen3-VL 235B is cheaper in all three workloads, by 27–32%. The table above uses list prices: MiniMax M3 $0.39 / 1M tokens input, $1.56 / 1M tokens output; Qwen3-VL 235B $0.26 / 1M tokens input, $1.15 / 1M tokens output.
How do I switch between them?
They are on the same endpoint and key. Change the model field from MiniMaxAI/MiniMax-M3 to Qwen/Qwen3-VL-235B-A22B-Instruct and nothing else.
Why does MiniMax M3 use more output tokens than the visible answer?
It is a reasoning model: it thinks before it answers and the thinking is billed as output tokens. Budget for that in output-heavy workloads, and set max_tokens to a few hundred or more so the answer is not cut off.
Spotted a price or limit that has changed? Tell us and we will re-check the provider page.