Qwen3.8 Max vs Hy3
At a Glance
| Compare |  Qwen3.8 MaxQwen |  Hy3Tencent |
|---|---|---|
| Intelligence, Cost, and Efficiency | ||
| IntelligenceHigher is better · MM Intelligence v2.5 | #10 of 4679.2 score · 2/3 sources · provisional · missing ARC-AGI-2 · full-core range 52.8–86.1 | UnrankedNot in the 46-model eligible cohort |
| CostLower is better · Published-token output estimate | #23 of 44$0.107 per LiveBench case | UnrankedNot in the 44-model eligible cohort |
| EfficiencyHigher is better · MM Efficiency v1.5 | #6 of 3863.2 score · 2/3 sources · provisional · missing ARC-AGI-2 · full-core range 50.0–66.7 | UnrankedNot in the 38-model eligible cohort |
| Pricing and Limits | ||
| Input priceFrom · USD / 1M tokens | $1.65Deepinfra ↗ · Sep 22, 2026 | $0.13Deepinfra ↗ · Sep 22, 2026 |
| Output priceFrom · USD / 1M tokens | $4.951Deepinfra ↗ · Sep 22, 2026 | $0.528Openrouter ↗ · Sep 22, 2026 |
| Context windowMaximum documented tokens | 1,000K | 262K |
| Model facts checked | Aug 29, 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 | Qwen3.8-Max | Hy3 |
|---|---|---|
| LMArena Agent Arenaagent-2026-09-15-d25aabda0010 · outcome_score · leader | 3.30100% of row best · score · Qwen3.8 Max; 95% CI [2.46604139, 4.14181105]; sessions 31489; observations 2502655; rank 17 | -5.2392% 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 · leader | 1,480.55100% of row best · rating · qwen3.8-max; 95% CI [1474.79040793, 1486.31846927]; votes 16670; rank 14 | 1,440.5897% of row best · rating · hy3; 95% CI [1433.39024152, 1447.77894384]; votes 8047; rank 65 |
| Overall ResultCounted from the protocol-matched rows above | 2 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 | Qwen3.8-Max | Hy3 |
|---|---|---|
| Developer | Qwen | Tencent |
| Family | Qwen3 8 Max | Hy3 |
| Model | Qwen3.8-Max | Hy3 |
| Version | Qwen3.8-Max | Hy3 |
| Lifecycle | active | active |
| Released | Unknown | Unknown |
| Knowledge cutoff | Unknown | Unknown |
| Input modalities | Text, Image, Video | Text |
| Output modalities | Text | Text |
| Context window | 1,000K | 262K |
| Total parameters | 2.4T | 298.8B |
| Active parameters | Unknown | 21B |
| License | Unknown | apache-2.0 |
| Open weights | No | Yes |
| API available | Yes | Yes |
| Self-hostable | No | Yes |
| Provider access | Alibaba Cloud Model Studio (Standard), Deepinfra (Standard), Fireworks Ai (Standard), Openrouter (Standard) | Deepinfra (Standard), Hugging Face (Standard), Openrouter (Standard) |
| Capabilities | agents, chat, reasoning, structured_outputs, tools, vision | chat, generation, tools |
Qwen3.8 Max Capabilities
Hy3 Capabilities
Related Comparisons
All text comparisons →| A | Pair | B | Context |
|---|---|---|---|
 GPT-6 AstraOpenAI | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
Claude Fable 5.1Anthropic | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
 Gemini 3.1 ProGoogle DeepMind | vs | Qwen3.8-MaxQwen | cross-developer peerstext |
 DeepSeek-V4.1-FlashDeepSeek | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
Qwen3.8-MaxQwen | vs | Grok 4.6xAI | cross-developer peersimage, text |
 Kimi-K3Moonshot AI | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
 MiniMax-M3MiniMax | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
Qwen3.8-MaxQwen | vs |  GLM-5.3Z.ai | cross-developer peerstext |
Qwen3.8-MaxQwen | vs | Hy4 previewTencent | cross-developer peerstext |
Seed 2.1 ProByteDance Seed | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text, video |
 Mistral Large 3Mistral AI | vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
| vs | Qwen3.8-MaxQwen | cross-developer peersimage, text |
Primary Evidence
Sources and Freshness
Questions
Qwen3.8 Max vs Hy3 FAQs
Is Qwen3.8 Max or Hy3 better for coding?+
This comparison does not currently contain a protocol-matched coding benchmark for both Qwen3.8 Max and Hy3, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.
Which is cheaper, Qwen3.8 Max or Hy3?+
Qwen3.8 Max is $1.65 and Hy3 is $0.13 per million tokens, so Hy3 is cheaper on this metric. Qwen3.8 Max is $4.951 and Hy3 is $0.528 per million tokens, so Hy3 is cheaper on this metric.
Which has a larger context window, Qwen3.8 Max or Hy3?+
Qwen3.8 Max has the larger sourced context window. Qwen3.8 Max supports 1,000K and Hy3 supports 262K.
Which performs better in benchmarks, Qwen3.8 Max or Hy3?+
There is no overall benchmark winner: An overall winner requires at least two decisive benchmarks from at least two original publishers.
Can Qwen3.8 Max or Hy3 be self-hosted?+
Hy3 is the only model in this pair currently marked as self-hostable. Qwen3.8 Max is not marked open weight; Hy3 is open weight.
Can Qwen3.8 Max and Hy3 understand images?+
Qwen3.8 Max 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, Qwen3.8 Max or Hy3?+
Neither has a larger sourced maximum output. Qwen3.8 Max is 131K and Hy3 is —.
Do Qwen3.8 Max and Hy3 support reasoning and tool use?+
Qwen3.8 Max: reasoning, tool calling, and image input. Hy3: tool calling. Feature support does not establish relative quality.
Which is available from more inference providers, Qwen3.8 Max or Hy3?+
Qwen3.8 Max has 4 sourced provider routes; Hy3 has 3, so Qwen3.8 Max has broader tracked availability.
Which offers better value, Qwen3.8 Max 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.