GPT-5.6 Luna vs Qwen3.5-27B
Side-by-side comparison of API pricing, specs, benchmarks, and capabilities
| Specification |
by OpenAI
|
by Alibaba
|
|---|---|---|
| Overview | ||
| Status | Current mid tier | Current mid tier Open weights |
| Released | Jul 9, 2026 | Feb 25, 2026 |
| Pricing per million tokens | ||
| Input | $0.200/Mtok | $0.300/Mtok |
| Output | $1.20/Mtok | $2.40/Mtok |
| Blended avg | $0.450/Mtok | $0.825/Mtok |
| Cached input | $0.020/Mtok | — |
| Price basis | — | Alibaba Model Studio International list price, single 0<Token≤256K tier, from the vendor's open-source Qwen table. Thinking-mode output is billed at the same rate as non-thinking. No context-cache or batch rate is printed for this SKU, so cached input is left unset rather than assumed. Gateway resellers quote lower routed rates ($0.195/$1.56). This SKU's row carries no discount label, so the standard list rate is the one tracked here — the same basis as every other Alibaba row on this site. |
| Specifications | ||
| Context window | 1M tokens | 262K tokens |
| Parameters | Proprietary | 27B |
| Speed (TPS) | — | — |
| Modalities | ||
| Input |
textimage
|
text
|
| Benchmarks sources: Vellum LLM Leaderboard / Artificial Analysis | ||
| Independent composite scores measured for both models — each on its own scale | ||
| AA-Omniscience Index (−100–100) | -10.3 | -44 |
| AA Intelligence Index | 52.3 | — |
| AA Agentic Index | 46.9 | — |
| GDPval-AA v2 | 1578.3 ELO | — |
| LMSYS Chatbot Arena | — | 1408 ELO |
| Per-benchmark results published for GPT-5.6 Luna only — no independent per-benchmark scores exist for Qwen3.5-27B yet | ||
| Avg benchmark score | 90.7 | — |
| Perf / dollar | 201.6 | — |
| GPQA Diamond | 92.3 | — |
| SWE-Bench Verified | 93 | — |
| Terminal-Bench 2.1 | 84.3 | — |
| HumanEval | 93 | — |
| Providers | ||
| Available from |
OpenAI — $0.200/$1.20/Mtok
|
Alibaba — $0.300/$2.40/Mtok
|
Cost at scale
1M tokens · 50/50 input/output| Volume | GPT-5.6 Luna | Qwen3.5-27B | Savings |
|---|---|---|---|
| 1M tokens | $0.45 | $0.83 | $0.37 (44.8%) |
| 10M tokens | $4.5 | $8.25 | $3.75 (45.5%) |
| 100M tokens | $45 | $82.5 | $37.5 (45.5%) |
| 1000M tokens | $450 | $825 | $375 (45.5%) |
When to pick which
distilled from the pricing and spec data above- cost dominates: $0.450/Mtok blended vs $0.825 — 45.5% less on the same 50/50 token mix
- your workload re-reads context (agents, RAG, long chats): cached input costs $0.020/Mtok — 90% off its list input price, a discount Qwen3.5-27B doesn't offer
- you need the longer context: 1M tokens vs 262K (4×)
- you want open weights — self-host it, fine-tune it, or exit the API entirely; GPT-5.6 Luna is closed
Price movement
recorded changes since we began tracking each model · last checked Aug 18, 2026| Model | Changes | Latest move | Old → new $/M |
|---|---|---|---|
| GPT-5.6 Luna | 1 | price cut on Aug 1, 2026 | $1/$6 → $0.2/$1.2/Mtok |
| Qwen3.5-27B | 0 | — no move | Stable since we began tracking — no change recorded |
The cheaper option here — GPT-5.6 Luna costs $0.450/Mtok blended on OpenAI.
Summary
GPT-5.6 Luna by OpenAI costs $0.200/Mtok input and $1.20/Mtok output, with a 1M-token context window. It supports text, image input.
Qwen3.5-27B by Alibaba costs $0.300/Mtok input and $2.40/Mtok output, with a 262K-token context window. It supports text input.
On a blended cost basis, GPT-5.6 Luna is 45.5% cheaper than Qwen3.5-27B.It also has a larger context window.
The two aren't directly comparable on average benchmark score: GPT-5.6 Luna has published per-benchmark results, while Qwen3.5-27B does not yet — it is measured today on independent composites (see the table above).
Note: Pricing is per million tokens. Actual costs vary with usage patterns, prompt caching, and batch discounts. Always verify against official provider pricing pages.