Cheapest MiniMax Model (2026)

The cheapest MiniMax model on a per-token basis is minimax-01 at $0.20 input and $1.10 output per million tokens (measured via OpenRouter), but m2.7 at $0.24/$0.96 wins for output-heavy work because its output rate is the lowest in the lineup. Which one is actually cheapest for you depends on whether your workload reads more than it writes.

The headline "cheapest" number is a trap on the MiniMax lineup, because the five tiers do not rank the same way on input as they do on output. minimax-01 has the lowest input rate but a mid-pack output rate. m2.7 has a slightly higher input rate but the lowest output rate of all. So the answer flips depending on your token mix, and most teams pick the wrong one by reading only the first column of a price table.

Cheapest MiniMax model, ranked (verified 2026-06)

Here is the full ladder, sorted by input price. All five rates were read from the live OpenRouter catalog on 2026-06-26 and are the OpenRouter-routed values, not native MiniMax billing. Prices are per million tokens, USD.

| Rank (by input) | Model | Input $/1M | Output $/1M | Cheapest for | |---|---|---|---| | 1 | minimax-01 | $0.20 | $1.10 | Input-heavy: long-context reads, RAG | | 2 | minimax-m2.7 | $0.24 | $0.96 | Output-heavy: generation, agents | | 3 | minimax-m2 | $0.255 | $1.00 | Balanced general use | | 4 | minimax-m3 | $0.30 | $1.20 | Newest tier, promotional rate | | 5 | minimax-m1 | $0.40 | $2.20 | Legacy; avoid for cost |

The practical upshot: if your prompts are long and your completions are short, minimax-01 is your cheapest option. If you generate a lot of text per call, m2.7's $0.96 output rate beats every other tier and the 4-cent input premium over minimax-01 rarely matters. minimax-m1 sits at the bottom of the value ranking. At $0.40/$2.20 it is the most expensive tier on both columns, so reach for it only if a specific capability justifies the premium.

A definition worth pinning down first. The per-token rate is a unit price that bills separately for the tokens you send (input) and the tokens the model generates (output), which is why a single "cheapest" label cannot capture a model whose two rates rank differently against its siblings.

According to MiniMax, the official platform serves these models internationally and lists its own rate card, which does not always match the OpenRouter-routed numbers above. The clearest example is m2: official MiniMax pricing puts it at $0.30 input and $1.20 output, while the OpenRouter route we measured bills $0.255/$1.00. The OpenRouter path is the cheaper of the two for that model today. These official figures need native-key re-verification, so treat them as sourced rather than first-hand.

The M3 permanent 50%-off note most tables miss

According to MiniMax, the M3 rate of $0.30 input and $1.20 output is a permanent 50%-off the raw $0.60/$2.40 card for prompts up to 512K input tokens. That promotional structure is why M3 sits mid-ladder rather than near the top despite being the newest tier. The catch lives above the line: inputs over 512K tokens fall back to the full $0.60/$2.40 rate, and that larger-context tier is access-limited and routed through MiniMax sales rather than self-serve. So the cheap M3 number you see assumes you stay under 512K input per request, which most chat and agent workloads do comfortably.

This is the kind of nuance that breaks a naive "cheapest model" comparison. M3 looks like a flat $0.30/$1.20 tier in a spreadsheet, but it is really a discounted band with a cliff at 512K input. Budget against the band you actually use.

We measured three tiers live via OpenRouter

Documentation gives you rate cards. We wanted billed numbers, so we routed real calls through OpenRouter on 2026-06-26 and recorded what each tier actually cost per request. These are OpenRouter-routed measurements, not native MiniMax billing, and we disclose that because the routing layer sets the price you see here.

The minimax-m2 general call billed $0.000209945 for 39 input and 200 output tokens, served in 5.08 seconds. The minimax-m1 call, on a near-identical 39-token prompt, billed $0.0004138 for fewer output tokens (181), which is almost double the m2 cost for less generated text. That single comparison is the whole value argument against m1: you pay more and get less per dollar. The m3 call billed $0.00027114 and was served by the dated build minimax-m3-20260531, confirming the promotional tier is the one routed by default.

One honest caveat on these completions. Our general calls hit the 200-token output cap, so the m2 and m3 completion_tokens of 200 are capped, not natural stopping points. We treat them as a fixed-output cost probe, not a measure of how verbose each model is.

How this compares to DeepSeek

For teams cross-shopping Chinese models, DeepSeek is the obvious price anchor. According to DeepSeek API Docs, DeepSeek V4-Flash bills $0.14 input and $0.28 output per million tokens at the official endpoint, which undercuts even minimax-01's $0.20 input and sits far below every MiniMax output rate. The catch is the comparison is not apples to apples: the DeepSeek figure is official-endpoint billing while every MiniMax number here is OpenRouter-routed, so part of the gap is the routing layer's margin, not the underlying model. Still, on raw rate card alone, DeepSeek V4-Flash is cheaper than the cheapest MiniMax tier. For the full DeepSeek breakdown, see our DeepSeek API pricing hub. The MiniMax case is rarely about beating DeepSeek on price; it is about a specific capability or output quality at a still-low rate.

On context window, we deliberately leave one number out. MiniMax-M1's advertised context has been quoted inconsistently across sources, so we are not stating a figure here. If long-context capacity is your deciding factor, verify the current limit in the platform.minimax.io console for the exact model you plan to call, because the tier you provision determines it.

For the full cluster view, the MiniMax API pricing hub ranks every tier with the same evidence and links each focused breakdown.

FAQ

What is the cheapest MiniMax model? On a per-token basis, minimax-01 at $0.20 input and $1.10 output per million tokens is the cheapest, measured via OpenRouter on 2026-06-26. But m2.7 at $0.24/$0.96 is cheaper for output-heavy workloads because its output rate is the lowest in the lineup. The right answer depends on your input-to-output token ratio.

Is minimax-01 always cheaper than m2.7? No. minimax-01 wins on input ($0.20 vs $0.24) but m2.7 wins on output ($0.96 vs $1.10). For a workload that generates far more tokens than it reads, m2.7 is the cheaper total. For long-context reads with short completions, minimax-01 wins.

Why is M3 only $0.30 when its raw rate is $0.60? According to MiniMax pricing, the $0.30/$1.20 M3 rate is a permanent 50%-off promotion for inputs up to 512K tokens. Above 512K input, it reverts to the full $0.60/$2.40 rate on an access-limited tier routed through sales. These official figures need native-key re-verification.

How much did a real MiniMax call cost in your test? A minimax-m2 call with 39 input and 200 output tokens billed $0.000209945 in 5.08 seconds, measured via OpenRouter on 2026-06-26. The same-size minimax-m1 prompt billed $0.0004138 for fewer output tokens, nearly double the cost.

Is MiniMax cheaper than DeepSeek? No. DeepSeek V4-Flash at $0.14/$0.28 (official endpoint) undercuts even the cheapest MiniMax tier. But the MiniMax numbers here are OpenRouter-routed, so part of the gap is the routing margin rather than the model itself.

---

This is part of the MiniMax API pricing hub, which ranks every tier with the same evidence.

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

<!-- METADATA { "title": "The Cheapest MiniMax API Tier, Ranked (2026)", "slug": "cheapest-minimax-model", "meta_description": "Ranked: minimax-01 $0.20/$1.10, m2.7 $0.24/$0.96, m2 $0.255/$1.00, m3 $0.30/$1.20, m1 $0.40/$2.20 (via OpenRouter). Per-token nuance + M3 50%-off note. 2026-06.", "focus_keyword": "cheapest minimax model", "secondary_keywords": ["cheapest minimax api", "minimax api pricing ranked", "minimax-01 price", "minimax m2.7 cost", "minimax m3 pricing"], "tags": ["MiniMax", "API Pricing", "Model Comparison"], "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": "What is the cheapest MiniMax model?", "a": "On a per-token basis, minimax-01 at $0.20 input and $1.10 output per million tokens is the cheapest, measured via OpenRouter on 2026-06-26. But m2.7 at $0.24/$0.96 is cheaper for output-heavy workloads because its output rate is the lowest in the lineup."}, {"q": "Is minimax-01 always cheaper than m2.7?", "a": "No. minimax-01 wins on input ($0.20 vs $0.24) but m2.7 wins on output ($0.96 vs $1.10). For a workload that generates far more tokens than it reads, m2.7 is the cheaper total."}, {"q": "Why is M3 only $0.30 when its raw rate is $0.60?", "a": "According to MiniMax pricing, the $0.30/$1.20 M3 rate is a permanent 50%-off promotion for inputs up to 512K tokens. Above 512K input, it reverts to the full $0.60/$2.40 rate on an access-limited tier. These official figures need native-key re-verification."}, {"q": "How much did a real MiniMax call cost in your test?", "a": "A minimax-m2 call with 39 input and 200 output tokens billed $0.000209945 in 5.08 seconds, measured via OpenRouter on 2026-06-26. The same-size minimax-m1 prompt billed $0.0004138 for fewer output tokens, nearly double the cost."}, {"q": "Is MiniMax cheaper than DeepSeek?", "a": "No. DeepSeek V4-Flash at $0.14/$0.28 (official endpoint) undercuts even the cheapest MiniMax tier. But the MiniMax numbers here are OpenRouter-routed, so part of the gap is the routing margin rather than the model itself."} ], "external_links_used": [ {"url": "https://platform.minimax.io", "source_name": "MiniMax Platform", "claim": "Official international platform and rate card; M3 permanent 50%-off below 512K input; verify M1 context window in console"}, {"url": "https://api-docs.deepseek.com/quick_start/pricing/", "source_name": "DeepSeek API Docs – Pricing", "claim": "DeepSeek V4-Flash official endpoint pricing $0.14 input / $0.28 output per 1M tokens"} ], "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": "cross-cluster"} ], "first_hand_evidence": { "source": "minimax-api-pricing-evidence.json runs: minimax-m2_general, minimax-m1_general, minimax-m3_general", "measured": "minimax-m2: 39 in / 200 out, $0.000209945, 5.08s; minimax-m1: 39 in / 181 out, $0.0004138, 5.45s; minimax-m3 (served minimax-m3-20260531): 195 in / 200 out, $0.00027114, 5.20s. All OpenRouter-routed.", "captured": "2026-06-26" }, "images_status": "spec-only (not generated; FAL_API_KEY unset)", "images": [ {"position": "featured", "type": "generated", "prompt": "Clean editorial ranking chart of five MiniMax API tiers sorted by per-token price, showing minimax-01, m2.7, m2, m3, m1 with input and output rates as paired bars. Note that input-cheapest and output-cheapest are different models. Teal and amber palette. Minimal background. 16:9.", "alt": "Ranked bar chart of five MiniMax API tiers by per-token input and output price, showing minimax-01 cheapest on input and m2.7 cheapest on output"} ] } -->