MiniMax API Cost Calculator (2026)

To budget a MiniMax API workload, blend the input and output rates by your real token ratio: at MiniMax-M2's OpenRouter-routed $0.255 in / $1.00 out, a 1:2 input-to-output mix costs about $0.752 per million tokens, and M3's promo rate of $0.30 / $1.20 runs slightly higher. That blended number, not the headline input price, is what you should multiply your monthly token volume against. Below we walk through the formula, anchor it to a call we actually billed, and flag the caveats that trip up first-time budgeters.

The blended-cost formula (verified 2026-06)

A cost calculator for any LLM API is just one equation applied honestly. The trap is reading the input price as "the price." Output tokens are almost always more expensive and usually more numerous, so they dominate the bill.

The formula:

blended cost per 1M tokens =
  (input_share × input_rate) + (output_share × output_rate)

For MiniMax-M2 at the OpenRouter-routed $0.255 in / $1.00 out, a workload that is one-third input and two-thirds output works out to (0.333 × $0.255) + (0.667 × $1.00), or roughly $0.752 per million tokens. The practical upshot is that your input price barely moves the result. Output is doing most of the work, which is why a calculator that only shows you the input rate will under-budget you by a wide margin.

Here is how the two current MiniMax tiers compare on a 1:2 input-to-output mix, with DeepSeek's official V4-Flash rate alongside as a reference point.

According to MiniMax pricing, the M3 promotional rate of $0.30 in / $1.20 out is a permanent 50% discount off the $0.60 / $2.40 raw rate, applied to requests up to 512K input tokens. Treat the promo as the rate you budget against today, but flag it for re-verification because it is a discount that could change.

What a real billed call says (first-hand evidence)

A formula is only as trustworthy as the rate inside it, so we ran live calls rather than trusting a rate card alone. Every MiniMax figure here was measured via OpenRouter, because MiniMax is reached through OpenRouter in our setup rather than a native key.

When we called MiniMax-M2 with a short general-purpose prompt, the request used 39 input tokens and 200 output tokens and billed $0.000209945, returning in about 5.08 seconds. A second M2 call sized for a coding task used 38 input and 200 output tokens and billed $0.0002514. The M3 call (served as minimax-m3-20260531) used 195 input and 200 output tokens and billed $0.00027114.

One caveat matters for anyone scaling these numbers up: our general-purpose calls hit a 200-token output cap, so the completion_tokens=200 you see above is the ceiling we set, not the model running out of things to say. When you model a production workload that generates 800 or 1,500 output tokens per call, scale the output side accordingly. The per-call cost above is a unit-test of the rate, not a forecast of your average request.

Stepping from one call to a monthly budget

Once you trust the blended rate, projecting a month is arithmetic. Suppose a workload runs 500,000 calls a month, each averaging 300 input tokens and 600 output tokens. That is 150 million input tokens and 300 million output tokens.

On MiniMax-M2's OpenRouter-routed rates: input is 150M × $0.255/1M = $38.25, output is 300M × $1.00/1M = $300, for $338.25 total. The output line is roughly 89% of the bill. If you only sized against the input rate, you would have budgeted $38 and missed the real number by an order of magnitude. This is the single most common budgeting error we see, and it is why the blended formula leads this article.

According to MiniMax pricing, the same workload on the M3 promo tier ($0.30 / $1.20) would bill $45 input plus $360 output, or $405. The roughly 20% premium over M2 buys you the newer model; whether that is worth it depends on your quality testing, which is a separate question from cost.

How MiniMax sits against DeepSeek on cost

A blended-cost calculator is a comparison tool that lets you rank providers on the only axis that survives a procurement review: dollars per real workload. For a DeepSeek baseline, we anchored to an official-endpoint call rather than a routed one, so the comparison is honest about its asymmetry. Our DeepSeek V4-Flash coding call used 27 input and 86 output tokens and returned in 1.96 seconds, costed at the official $0.14 / $0.28 rate.

According to DeepSeek API Docs, V4-Flash bills $0.14 in / $0.28 out, which lands around $0.233 per million tokens on a 1:2 mix. That is roughly a third of MiniMax-M2's blended cost. The honest read is that DeepSeek is the cheaper line item, and MiniMax has to win on output quality or capability for a given task to justify the gap. We measured MiniMax via OpenRouter and DeepSeek on its official endpoint, so the two are not perfectly apples-to-apples; routing fees and rate-card differences both sit inside that comparison. For the full DeepSeek rate breakdown, see the DeepSeek API pricing hub.

A definition worth pinning down: blended cost is a single per-million-token rate that weights the separate input and output prices by the actual ratio of input to output tokens in a workload, so it differs from the headline input rate that vendors lead with.

For the full MiniMax rate card, tier notes, and the rest of this cluster, start at the MiniMax API pricing hub.

FAQ

How do I calculate MiniMax API cost per month? Multiply your monthly input tokens by the input rate and your output tokens by the output rate, then add them. On MiniMax-M2's OpenRouter-routed $0.255 / $1.00, a workload of 150M input and 300M output tokens bills $38.25 + $300 = $338.25. Output usually dominates, so size against it, not the input price.

What is the blended cost per million tokens for MiniMax-M2? On a 1:2 input-to-output ratio, MiniMax-M2 at the OpenRouter-routed $0.255 in / $1.00 out blends to roughly $0.752 per million tokens. Change the ratio and the number moves; a more output-heavy workload pushes it toward the $1.00 output rate.

Is MiniMax-M3 more expensive than M2? Yes, modestly. M3's promotional rate of $0.30 / $1.20 (measured via OpenRouter) is about 20% above M2's $0.255 / $1.00. According to MiniMax pricing, the M3 promo is a permanent 50% discount off a $0.60 / $2.40 raw rate for requests up to 512K input, so verify the current tier in your console before committing a budget.

Why is my per-call test cost lower than my projected bill? Our first-hand calls hit a 200-token output cap, so each billed only a fraction of a cent. Real production calls generate far more output tokens, and output is the expensive side. Scale the output token count to your actual average before forecasting.

Is MiniMax cheaper than DeepSeek? No. DeepSeek V4-Flash at the official $0.14 / $0.28 blends to about $0.233 per million tokens, roughly a third of MiniMax-M2's $0.752. MiniMax has to justify the gap on capability. Note we measured MiniMax via OpenRouter and DeepSeek on its official endpoint.

---

This is part of the MiniMax API pricing hub, which carries the full rate card and tier caveats.

Author: Kevin Fan, Customer Success Manager at China LLM Directory, specializing in Chinese LLM ecosystem pricing. Last verified: 2026-06-26.

<!-- METADATA { "title": "How to Calculate MiniMax API Costs (2026)", "slug": "minimax-api-cost-calculator", "meta_description": "Budget MiniMax API spend with the blended-cost formula: M2 at $0.255/$1.00 blends to ~$0.752/1M tokens. First-hand billed calls, M3 promo notes, 2026-06.", "focus_keyword": "minimax api cost calculator", "secondary_keywords": ["minimax api cost", "minimax blended cost", "minimax m2 pricing calculator", "how to calculate minimax api cost"], "tags": ["MiniMax", "API Pricing", "Cost Calculator"], "category": "Pricing", "cluster_id": "minimax-api-pricing", "cluster_role": "micro", "hub_slug": "minimax-api-pricing", "evidence_file": "clients/china-llm-aggregator/articles/minimax-api-pricing-evidence.json", "needs_native_reverify": true, "verified_until": "2026-09-24", "author_name": "Kevin Fan", "author_title": "Customer Success Manager", "author_linkedin": "", "author_expertise": ["Chinese LLM ecosystem", "AI infrastructure pricing", "model benchmarking", "cross-border AI compliance"], "faq_pairs": [ {"q": "How do I calculate MiniMax API cost per month?", "a": "Multiply monthly input tokens by the input rate and output tokens by the output rate, then add them. On MiniMax-M2's OpenRouter-routed $0.255 / $1.00, 150M input and 300M output tokens bills $38.25 + $300 = $338.25. Output usually dominates, so size against it."}, {"q": "What is the blended cost per million tokens for MiniMax-M2?", "a": "On a 1:2 input-to-output ratio, MiniMax-M2 at the OpenRouter-routed $0.255 in / $1.00 out blends to roughly $0.752 per million tokens. A more output-heavy workload pushes it toward the $1.00 output rate."}, {"q": "Is MiniMax-M3 more expensive than M2?", "a": "Yes, modestly. M3's promotional rate of $0.30 / $1.20 (measured via OpenRouter) is about 20% above M2's $0.255 / $1.00. According to MiniMax pricing, the M3 promo is a permanent 50% discount off a $0.60 / $2.40 raw rate for up to 512K input; verify in your console."}, {"q": "Why is my per-call test cost lower than my projected bill?", "a": "Our first-hand calls hit a 200-token output cap, so each billed a fraction of a cent. Real production calls generate far more output tokens, and output is the expensive side. Scale the output count to your actual average before forecasting."}, {"q": "Is MiniMax cheaper than DeepSeek?", "a": "No. DeepSeek V4-Flash at the official $0.14 / $0.28 blends to about $0.233 per million tokens, roughly a third of MiniMax-M2's $0.752. We measured MiniMax via OpenRouter and DeepSeek on its official endpoint."} ], "external_links_used": [ {"url": "https://platform.minimax.io", "source_name": "MiniMax pricing", "claim": "M3 promo $0.30/$1.20 is a permanent 50% off the $0.60/$2.40 raw rate for up to 512K input; needs_native_reverify"}, {"url": "https://api-docs.deepseek.com/quick_start/pricing/", "source_name": "DeepSeek API Docs", "claim": "DeepSeek V4-Flash bills $0.14 in / $0.28 out"} ], "internal_links_used": [ {"url": "/blog/minimax-api-pricing/", "anchor_text": "MiniMax API pricing hub", "type": "hub"}, {"url": "/blog/deepseek-api-pricing/", "anchor_text": "DeepSeek API pricing hub", "type": "external-cluster-hub"} ], "first_hand_evidence": { "source": "minimax-api-pricing-evidence.json runs minimax-m2_general, minimax-m2_coding, minimax-m3_general, deepseek_v4flash_coding_anchor", "measured": "M2 general 39 in/200 out $0.000209945 5.08s; M2 coding 38 in/200 out $0.0002514 2.95s; M3 general (minimax-m3-20260531) 195 in/200 out $0.00027114 5.20s; DeepSeek V4-Flash 27 in/86 out 1.96s official. MiniMax measured via OpenRouter; 200-token output cap.", "captured": "2026-06-26" }, "images_status": "spec-only (not generated; FAL_API_KEY unset)", "images": [ {"position": "featured", "type": "generated", "prompt": "Clean editorial diagram of the blended-cost formula for MiniMax API, showing input rate and output rate weighted by token ratio into a single per-million-token number, with MiniMax-M2 $0.255/$1.00 annotated. Teal and amber palette. Minimal background. 16:9.", "alt": "Diagram of the MiniMax API blended-cost formula weighting input and output rates by token ratio, annotated with MiniMax-M2 OpenRouter rates of $0.255 input and $1.00 output"} ] } -->