Gemini 3 Flash Preview vs MiniMax M2.5
Gemini 3 Flash Preview (Google, released 2025-12-17) and MiniMax M2.5 (MiniMax, released 2026-02-12) target overlapping use cases — agentic coding, multi-turn chat, tool use — but differ sharply on price, context, and modality. Gemini 3 Flash handles text, image, file, audio, and video input with a 1,048,576-token context window; MiniMax M2.5 is text-only with a 204,800-token window but costs a fraction as much per token. Both are available on OpenKey with one API key and a flat 3% fee on top of provider list pricing.
Spec vs spec
| Spec | Gemini 3 Flash Preview | MiniMax M2.5 |
|---|---|---|
| Context window | 1.0M | 205K |
| Max output | 66K | 197K |
| Input modalities | text, image, file, audio, video | text |
| Output modalities | text | text |
| Released | Dec 17, 2025 | Feb 12, 2026 |
| Reasoning | optional | always on |
Pricing
Per 1M tokens. Provider price plus the flat 3% fee — the sum is what you pay.
google/gemini-3-flash-preview
Input · 1M tokens
$0.500 + 3%$0.515
Output · 1M tokens
$3.00 + 3%$3.09
Cache read · 1M tokens
$0.050 + 3%$0.052
Cache write · 1M tokens
$0.083 + 3%$0.086
FEE — FLAT, EVERY MODEL3%
minimax/minimax-m2.5
Input · 1M tokens
$0.120 + 3%$0.124
Output · 1M tokens
$0.480 + 3%$0.494
FEE — FLAT, EVERY MODEL3%
One workload, priced on both
10M input + 2M output tokens at each model's price, flat 3% fee included.
google/gemini-3-flash-preview
$11.33
$11.00 provider + 3%
minimax/minimax-m2.5Cheaper
$2.22
$2.16 provider + 3%
Benchmarks
Design Arena categories where both models have results. Higher Elo and lower rank win.
| Gemini 3 Flash Preview | MiniMax M2.5 | |||
|---|---|---|---|---|
| Category | Elo | Rank | Elo | Rank |
| 3D | 1261 | #29 | 1246 | #31 |
| Code | 1239 | #33 | 1256 | #29 |
| Game dev | 1232 | #38 | 1242 | #33 |
| Websites | 1239 | #33 | 1265 | #26 |
Head-to-head preference voting. How we filter and rank
Pricing math on a real workload
Provider list price for Gemini 3 Flash Preview is $0.50/M input tokens and $3.00/M output tokens; on OpenKey that's $0.515/M input and $3.09/M output after the 3% fee ($0.50 × 1.03, $3.00 × 1.03). MiniMax M2.5 lists at $0.12/M input and $0.48/M output, or $0.1236/M and $0.4944/M on OpenKey.
Run the numbers on a 10M-input / 2M-output workload: Gemini 3 Flash Preview costs $11.00, MiniMax M2.5 costs $2.16 — a 5.1x gap. The input price ratio alone is 4.17x. If you're running high-volume agent loops or batch jobs, that difference compounds fast.
Coding and agent benchmarks
Both models have Design Arena scores in the models arena. On codecategories, MiniMax M2.5 scores elo 1256 (rank 29, 56.8% win rate) versus Gemini 3 Flash Preview's elo 1239 (rank 33, 57.6% win rate) — MiniMax edges ahead on rank and elo despite a slightly lower win rate. On website, MiniMax M2.5 leads with elo 1265 (rank 26) against Gemini's elo 1239 (rank 33). On 3D, Gemini 3 Flash Preview actually comes out ahead: elo 1261 (rank 29) versus MiniMax M2.5's elo 1246 (rank 31).
Gemini 3 Flash Preview also has agent-specific arena data MiniMax lacks in this dataset — fullstack (elo 1130, rank 17), webapps (elo 1185, rank 17), and godotgamedev (elo 1218, rank 8) — useful if you need agentic tool-calling benchmarks specifically.
Context and modality
Gemini 3 Flash Preview's 1,048,576-token context is 5.12x larger than MiniMax M2.5's 204,800 tokens — the context_ratio the numbers confirm. That matters for ingesting long codebases, transcripts, or multi-document research in a single call. Gemini also accepts image, file, audio, and video input, while MiniMax M2.5 is text-in/text-out only.
MiniMax M2.5 compensates with a larger max output: 196,608 tokens versus Gemini's 65,535. If your workload is about generating long structured output — full applications, long-form docs — rather than ingesting large context, that ceiling matters more than the input window.
Reasoning and tool support
MiniMax M2.5 has mandatory reasoning — it always runs a reasoning pass, with reasoning_effort as a supported parameter alongside a wider parameter set (logprobs, top_k, min_p, penalties). Gemini 3 Flash Preview's reasoning is optional, with four effort levels (high, medium, low, minimal) and medium as default, so you can dial cost and latency per request.
Both support tool_choice, tools, and structured_outputs, so either works for function-calling agent setups. If you need cheap, fast calls without paying for a reasoning pass every time, Gemini's optional reasoning gives more control.
Which model for which job
| Use case | Pick | Why |
|---|---|---|
| High-volume coding agent loop | MiniMax M2.5 | $2.16 vs $11.00 on a 10M-in/2M-out workload — 5.1x cheaper |
| Long-document or codebase ingestion | Gemini 3 Flash Preview | 1,048,576-token context vs 204,800 — 5.12x larger |
| Multimodal input (image/audio/video/file) | Gemini 3 Flash Preview | MiniMax M2.5 is text-only input |
| Long structured output generation | MiniMax M2.5 | 196,608 max output tokens vs Gemini's 65,535 |
| Website UI generation | MiniMax M2.5 | elo 1265, rank 26 on Design Arena website category vs Gemini's elo 1239, rank 33 |
| 3D generation tasks | Gemini 3 Flash Preview | elo 1261, rank 29 vs MiniMax's elo 1246, rank 31 |
Questions
- Which model is cheaper for a typical coding workload?
- MiniMax M2.5, by a wide margin. On a 10M-input/2M-output workload it costs $2.16 on OpenKey versus $11.00 for Gemini 3 Flash Preview — the input price ratio alone is 4.17x in MiniMax's favor.
- Does MiniMax M2.5 beat Gemini 3 Flash Preview on every benchmark?
- No. MiniMax M2.5 leads on codecategories (elo 1256 vs 1239) and website (elo 1265 vs 1239), but Gemini 3 Flash Preview ranks higher on 3D — elo 1261 (rank 29) versus MiniMax's elo 1246 (rank 31).
- Which model has a bigger context window?
- Gemini 3 Flash Preview, at 1,048,576 tokens versus MiniMax M2.5's 204,800 tokens — a 5.12x difference. MiniMax compensates with a larger max output ceiling of 196,608 tokens versus Gemini's 65,535.
- Can I use both models through one API key?
- Yes. Both Gemini 3 Flash Preview and MiniMax M2.5 are available on OpenKey with a single key across all 329 models on the platform, with pricing set at provider list price plus a flat 3% fee.