Qwen3.8-Max is Alibaba's current flagship on Model Studio, released August 3, 2026. It costs $2.00 per 1M input tokens and $6.00 per 1M output tokens with a 1M context window, and it's cheaper than the Qwen3.7-Max it replaces.
Qwen3.8-Max costs $2.00 input / $6.00 output per 1M tokens, with cache hits at $0.17. That undercuts the previous flagship, Qwen3.7-Max ($2.50 / $7.50), on both sides, so there's little reason to start new work on the older model.
| Qwen3.8-Max | |
|---|---|
| Provider | Alibaba |
| Tier | Flagship |
| Input / 1M tokens | $2.00 |
| Output / 1M tokens | $6.00 |
| Cache hit / 1M | $0.17 |
| Cache write / 1M | — |
| Context window | 1M |
| Notes | New Alibaba flagship (Aug 3, 2026); 2.4T MoE |
| Model | Support chatbot 10,000 conversations × (2,000 input + 400 output tokens) | Document Q&A 1,000 questions × (30,000 input + 800 output tokens) | Coding agent session 4M input tokens (90% cache hits) + 150,000 output tokens |
|---|---|---|---|
| Qwen3.8-Max | $64.00 | $64.80 | $2.31 |
| Grok 4.7 | $64.00 | $64.80 | $3.50 |
| Gemini 4 Argon | $80.00 | $68.00 | $2.66 |
| Kimi K3.1 Pro | $80.00 | $68.00 | $3.02 |
Standard rates per 1M tokens from data on this site. The coding-agent row assumes 90% of input is read from the prompt cache (full input price where a model has no cache rate). Cache-write fees, batch discounts and long-context surcharges listed in the notes are not included. Updated November 2, 2026.
| Model | Tier | Input / 1M | Output / 1M | Cache hit / 1M | Context | Notes |
|---|---|---|---|---|---|---|
| Qwen3.8-Max | Flagship | $2.00 | $6.00 | $0.17 | 1M | New Alibaba flagship (Aug 3, 2026); 2.4T MoE |
| Qwen3.7-Max | Mid-tier | $2.50 | $7.50 | $0.25 | 1M | Prior Qwen flagship; strong multilingual |
| Qwen3.7-Plus | Budget | $0.40 | $1.60 | $0.04 | 1M | Balanced Qwen workhorse; base tier list price shown (0-256K tokens); 256K-1M tokens: $1.20/$4.80; both bands currently carry a limited-time 20% discount |
| Model | Tier | Input / 1M | Output / 1M | Cache hit / 1M | Context | Notes |
|---|---|---|---|---|---|---|
| Grok 4.7 | Flagship | $2.00 | $6.00 | $0.50 | 500K | New flagship (Sep 21, 2026); larger base model, replaces 4.6 at the same price; ≥200K tokens: $4/$12, cache $1.00 |
| Gemini 4 Argon | Flagship | $2.00 | $10.00 | $0.10 | 1M | New top frontier model, generally available Oct 13, 2026 for AI Pro/Ultra subscribers and via the API (model id gemini-4-argon); long-decode generation up to 1M tokens; intro price thru Nov 30, 2026, then $4/$20 (cache read $0.20); not available on the free tier |
| Kimi K3.1 Pro | Flagship | $2.00 | $10.00 | $0.20 | 1M | Moonshot's current flagship, a further post-trained K3 at a lower price; cache write $2 (5-min TTL) or $4 (1-h TTL); Highspeed variant $4/$20 |
| Gemini 3.1 Pro | Flagship | $2.00 | $12.00 | $0.20 | 1M | Frontier multimodal (>200K: $4/$18, cache $0.40) |
| Kimi K3 | Flagship | $3.00 | $15.00 | $0.30 | 1M | Previous-generation flagship, superseded by Kimi K3.1 Pro but still sold; open weights; cache write $3 (5-min TTL) or $6 (1-h TTL) |
Alibaba prices Qwen3.8-Max on a single token tier covering every request up to 1M input tokens, so a 600K-token prompt is billed at the same per-token rate as a 6K one. That's different from Qwen3.7-Plus, which charges a higher rate once a request passes 256K input tokens. If your workload regularly sends very long documents, compare the two on your real prompt sizes rather than on headline rates. Both thinking and non-thinking modes are offered under the same model ID and price, but thinking mode produces more output tokens per answer, and output is the expensive side at $6.00.
Model Studio supports implicit caching (automatic) and explicit caching (you create the cache) for Qwen3.8-Max. Alibaba's context-cache docs say this model's cache-hit prices are not the usual 20% (implicit) and 10% (explicit) of the input rate that apply to Qwen3.7 models, and point to the console for its rates; the site's figure is $0.17. Explicit cache creation is billed at 125% of the input price, prompts shorter than 1,024 tokens can't be cached, and an explicit cache lives five minutes, reset by each hit. Batch and cache discounts can't be combined on the same request.
In the Singapore (international) region, the qwen3.8-max model includes a free quota of 1M tokens, valid for 90 days from Model Studio activation or the model's release, whichever is later. Free quotas aren't offered in other regions. Use the alias qwen3.8-max to track the latest version, or pin the snapshot qwen3.8-max-0902 if you need repeatable behavior for evaluations.
$2.00 per 1M input tokens and $6.00 per 1M output tokens, with cached input at $0.17, on a single price tier up to 1M input tokens.
Yes. Qwen3.7-Max is $2.50 input and $7.50 output per 1M tokens, so the newer flagship costs less on both input and output.
In the Singapore region, qwen3.8-max (and its qwen3.8-max-0902 snapshot) comes with a 1M-token free quota, valid for 90 days from Model Studio activation or the model's release, whichever is later. There is no ongoing free tier.
Yes. Alibaba lists qwen3.8-max with both thinking and non-thinking modes at the same token prices; thinking mode simply uses more output tokens.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.