MiniMax M3 vs Hy3
At a Glance
| Compare |  MiniMax M3MiniMax |  Hy3Tencent |
|---|---|---|
| Intelligence, Cost, and Efficiency | ||
| IntelligenceHigher is better · MM Intelligence v2.5 | #35 of 4632.6 score · 2/3 sources · provisional · missing ARC-AGI-2 · full-core range 21.8–55.1 | UnrankedNot in the 46-model eligible cohort |
| CostLower is better · Published-token output estimate | #3 of 44$0.017 per LiveBench case | UnrankedNot in the 44-model eligible cohort |
| EfficiencyHigher is better · MM Efficiency v1.5 | #10 of 3859.2 score · 2/3 sources · provisional · missing ARC-AGI-2 · full-core range 53.8–70.4 | UnrankedNot in the 38-model eligible cohort |
| Pricing and Limits | ||
| Input priceFrom · USD / 1M tokens | $0.28Deepinfra ↗ · Sep 22, 2026 | $0.13Deepinfra ↗ · Sep 22, 2026 |
| Output priceFrom · USD / 1M tokens | $1.10Deepinfra ↗ · Sep 22, 2026 | $0.528Openrouter ↗ · Sep 22, 2026 |
| Context windowMaximum documented tokens | 1,049K | 262K |
| Model facts checked | Aug 28, 2026View model evidence → | Aug 28, 2026View model evidence → |
Token prices are the lowest available sourced USD rates; input and output may use different providers. Cost ranking estimates output spend on LiveBench, not a full request bill. Ranking methodology →
Available Benchmarks
| Benchmark | MiniMax-M3 | Hy3 |
|---|---|---|
| LMArena Agent Arenaagent-2026-09-15-d25aabda0010 · outcome_score · statistical tie | -5.7699% of row best · score · Minimax M3; 95% CI [-6.59075506, -4.93697068]; sessions 39156; observations 2879059; rank 37 | -5.23100% of row best · score · Hy3; 95% CI [-6.39472360, -4.07137134]; sessions 27371; observations 992576; rank 35 |
| LMArena Text Arenatext-2026-09-13-d25aabda0010 · arena_rating · statistical tie | 1,433.50100% of row best · rating · minimax-m3; 95% CI [1429.31077607, 1437.68800840]; votes 48540; rank 82 | 1,440.58100% of row best · rating · hy3; 95% CI [1433.39024152, 1447.77894384]; votes 8047; rank 65 |
| Overall ResultCounted from the protocol-matched rows above · 2 ties | 0 benchmark winsNo overall winner | 0 benchmark winsNo overall winner |
Third-party benchmark Only like-for-like primary-publisher results are shown; raw scores, relative scores, configuration, and token spend remain visible.
Side-by-Side Facts
| Field | MiniMax-M3 | Hy3 |
|---|---|---|
| Developer | MiniMax | Tencent |
| Family | Minimax M3 | Hy3 |
| Model | MiniMax-M3 | Hy3 |
| Version | MiniMax-M3 | Hy3 |
| Lifecycle | active | active |
| Released | 2026-06-01 | Unknown |
| Knowledge cutoff | Unknown | Unknown |
| Input modalities | Text, Image | Text |
| Output modalities | Text | Text |
| Context window | 1,049K | 262K |
| Total parameters | 427B | 298.8B |
| Active parameters | 23B | 21B |
| License | other | apache-2.0 |
| Open weights | Yes | Yes |
| API available | Yes | Yes |
| Self-hostable | Yes | Yes |
| Provider access | Deepinfra (Standard), Fireworks Ai (Standard), Hugging Face (Standard), Openrouter (Standard), Together Ai (Standard) | Deepinfra (Standard), Hugging Face (Standard), Openrouter (Standard) |
| Capabilities | chat, generation, reasoning, tools | chat, generation, tools |
MiniMax M3 Capabilities
Hy3 Capabilities
Related Comparisons
All text comparisons →| A | Pair | B | Context |
|---|---|---|---|
MiniMax-M2.7MiniMax | vs | MiniMax-M3MiniMax | family variantstext |
MiniMax-M3MiniMax | vs |  GPT-6 AstraOpenAI | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs | Claude Fable 5.1Anthropic | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs |  Gemini 3.1 ProGoogle DeepMind | cross-developer peerstext |
MiniMax-M3MiniMax | vs |  DeepSeek-V4.1-FlashDeepSeek | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs | Grok 4.6xAI | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs |  Qwen3.8-MaxQwen | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs |  Kimi-K3Moonshot AI | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs |  GLM-5.3Z.ai | cross-developer peerstext |
MiniMax-M3MiniMax | vs | Hy4 previewTencent | cross-developer peerstext |
MiniMax-M3MiniMax | vs | Seed 2.1 ProByteDance Seed | cross-developer peersimage, text |
MiniMax-M3MiniMax | vs |  Mistral Large 3Mistral AI | cross-developer peersimage, text |
Primary Evidence
Sources and Freshness
Questions
MiniMax M3 vs Hy3 FAQs
Is MiniMax M3 or Hy3 better for coding?+
This comparison does not currently contain a protocol-matched coding benchmark for both MiniMax M3 and Hy3, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.
Which is cheaper, MiniMax M3 or Hy3?+
MiniMax M3 is $0.28 and Hy3 is $0.13 per million tokens, so Hy3 is cheaper on this metric. MiniMax M3 is $1.10 and Hy3 is $0.528 per million tokens, so Hy3 is cheaper on this metric.
Which has a larger context window, MiniMax M3 or Hy3?+
MiniMax M3 has the larger sourced context window. MiniMax M3 supports 1,049K and Hy3 supports 262K.
Which performs better in benchmarks, MiniMax M3 or Hy3?+
There is no overall benchmark winner: Neither model leads outside the score tolerance on the verified common benchmarks.
Can MiniMax M3 or Hy3 be self-hosted?+
Both models have the same recorded self-hosting status: supported. MiniMax M3 is open weight; Hy3 is open weight.
Can MiniMax M3 and Hy3 understand images?+
MiniMax M3 is documented with image input; Hy3 is not documented with image input. This reflects supported input modalities, not vision quality.
Which can generate longer answers, MiniMax M3 or Hy3?+
Neither has a larger sourced maximum output. MiniMax M3 is — and Hy3 is —.
Do MiniMax M3 and Hy3 support reasoning and tool use?+
MiniMax M3: reasoning, tool calling, and image input. Hy3: tool calling. Feature support does not establish relative quality.
Which is available from more inference providers, MiniMax M3 or Hy3?+
MiniMax M3 has 5 sourced provider routes; Hy3 has 3, so MiniMax M3 has broader tracked availability.
Which offers better value, MiniMax M3 or Hy3?+
There is no universal value winner. Compare the input and output prices above with the matched benchmark result for your workload: cheaper tokens can be offset by different quality, token usage, latency, or provider availability.