Qwen Qwen3.8 27B API Pricing and Speed
Qwen Qwen3.8 27B API pricing: $0.80 per 1M tokens for input and $4.00 per 1M tokens for output, from the official API price list of Groq. Official rate checked (source); OpenRouter is a separate channel below. Our measured latency and output speed are shown by route and country below. Published OpenRouter: Last checked · Source (data) · Methodology
OpenRouter: Amount worked out from the source page: $0.15 per 1M tokens, in USD · as shown in the United States on 9 Nov 2026 · the page does not say whether tax is included; we do not assume it is excluded · last price change seen on 31 Oct 2026.
Price changed 31 Oct 2026. output $1.25 → $1.875 (+50.0 %); input $0.039 → $0.15 (+284.6 %) per 1M tokens on OpenRouter. Confirmed by a later check on 1 Nov 2026.
Prices USD per 1M tokens, as listed
- $0.80Input per 1M tokens, from Groq Published
- $4.00Output per 1M tokens, from Groq Published
All prices by channel
| Price per 1M tokens | Groq direct | Through OpenRouter | Previous prices by channel |
|---|---|---|---|
| Input | $0.80 Published | $0.15 Published | OpenRouter: |
| Output | $4.00 Published | $1.875 Published | OpenRouter: |
| Cached input · read | UnknownUnknown | $0.0375 Published | — |
| Batch · input / output | — | $0.16 / $2.00 Published | — |
Input price from Groq, 24 Oct 2026 – 9 Nov 2026
24 Oct 2026$0.80 throughout9 Nov 2026
Input price through OpenRouter, 16 Oct 2026 – 9 Nov 2026
16 Oct 2026$0.24 → $0.15 (4 changes)9 Nov 2026
Context window1,024,000tokens Published
Max output235,929tokens Published
OpenRouter providers who serves this model through OpenRouter, at what price
OpenRouter sends a call without a pinned provider to one of these endpoints, and each endpoint has its own price. Prices are USD per 1M tokens, input / output.
Pinned min: input $0.035 · Darkbloom (darkbloom/fp4); output $1.05 · StreamLake (streamlake)
Unpinned range: input $0.035–$0.99, output $1.05–$3.20 across 23 providers (listing shows $0.15 / $1.875).
Price moves of 20% or more since the reading of 8 Nov 2026:
- Mancer 2 (mancer/fp8): $0.175 / $2.50 → $0.15 / $1.25 (input −14%, output −50%).
- PrimeIntellect (primeintellect/fp8): $0.021 / $1.159 → $0.45 / $3.20 (input +2043%, output +176%).
- Lowest-priced endpoint for the workload: Chutes (chutes/fp8) at $0.91 per 1,000 tasks, before PrimeIntellect (primeintellect/fp8) at $0.73.
Observed unpinned cost: 144 short requests sent without a pinned provider on 25 Oct 2026 – 9 Nov 2026 (16 dates), in each routing mode. Charged is the sum of the OpenRouter generation receipts for these requests; the listing price column is the price of the same input and output tokens at the listing price of that date.
| Routing mode | Requests | Dates | Served by | Charged | Listing price for the same tokens | Charged / listing | Lowest-priced endpoint answered |
|---|---|---|---|---|---|---|---|
| Default (no provider field) | 48 | 25 Oct 2026 – 9 Nov 2026 (16 dates) | 12 of 48 Mancer 2 (mancer/fp8), 12 of 48 ModelRun (modelrun/fp4), 10 of 48 Ionstream (ionstream/fp8), 3 of 48 PrimeIntellect (primeintellect/fp8), 2 of 48 AkashML (akashml/fp8), 2 of 48 Modal (modal), 2 of 48 Phala (phala), 1 of 48 DeepInfra (deepinfra/bf16), 1 of 48 Parasail (parasail/fp8), 1 of 48 Reka (reka/fp8), 1 of 48 Tenstorrent (tenstorrent/fp4), 1 of 48 Tenstorrent (tenstorrent/fp8) | $0.0037 | $0.0035 | 1.00× | 0 of 48 |
:floor | 48 | 25 Oct 2026 – 9 Nov 2026 (16 dates) | 26 of 48 Darkbloom (darkbloom/fp4), 15 of 48 PrimeIntellect (primeintellect/fp8), 4 of 48 Chutes (chutes/fp8), 3 of 48 Tenstorrent (tenstorrent/fp4) | $0.0020 | $0.0034 | 0.57× | 16 of 48 |
:nitro | 48 | 25 Oct 2026 – 9 Nov 2026 (16 dates) | 46 of 48 Cerebras (cerebras/fp16), 2 of 48 Nebius (nebius/fp4) | $0.0053 | $0.0034 | 1.45× | 0 of 48 |
Routing modes as documented by OpenRouter (OpenRouter provider routing, read 25 Oct 2026): the default request has no provider field; :floor and :nitro are suffixes of the model name. Served by names the endpoint that answered, by the provider name and endpoint tag of the receipt. Each request asks for one short line with an answer cap of 32 tokens. The counts are for these requests on these dates only.
Models with the widest provider price spread on OpenRouter
Request fragment that pins Darkbloom (darkbloom/fp4):
{"provider": {"order": ["darkbloom/fp4"], "allow_fallbacks": false}}Request fragment that pins StreamLake (streamlake):
{"provider": {"order": ["streamlake"], "allow_fallbacks": false}}With fallbacks turned off, OpenRouter does not move the call to another provider. A full endpoint tag pins that endpoint; the provider name alone covers all of its endpoints (OpenRouter provider routing).
All endpoints (23), by input price
| Provider | Endpoint tag | Precision | Input | Output | Context, tokens |
|---|---|---|---|---|---|
| Darkbloom | darkbloom/fp4 | fp4 | $0.035 | $1.50 | 262,144 |
| Chutes | chutes/fp8 | fp8 | $0.039 | $1.25 | 262,144 |
| Near AI | near-ai/fp8 | fp8 | $0.04 | $1.35 | 262,144 |
| Ionstream | ionstream/fp8 | fp8 | $0.04 | $1.95 | 262,144 |
| Modal | modal | Not stated | $0.04 | $2.30 | 262,144 |
| DekaLLM | dekallm | Not stated | $0.04 | $2.50 | 262,144 |
| Wafer | wafer/fp8 | fp8 | $0.06 | $2.30 | 262,144 |
| Mancer 2 | mancer/fp8 | fp8 | $0.15 | $1.25 | 262,144 |
| DeepInfra | deepinfra/bf16 | bf16 | $0.15 | $1.875 | 262,144 |
| Reka | reka/fp8 | fp8 | $0.15 | $2.50 | 262,144 |
| StreamLake | streamlake | Not stated | $0.175 | $1.05 | 1,024,000 |
| ModelRun | modelrun/fp4 | fp4 | $0.20 | $2.50 | 262,144 |
| Phala | phala | Not stated | $0.20 | $2.50 | 1,000,000 |
| AkashML | akashml/fp8 | fp8 | $0.225 | $1.98 | 262,144 |
| Parasail | parasail/fp8 | fp8 | $0.24 | $2.20 | 262,144 |
| CoreWeave | coreweave/fp8 | fp8 | $0.40 | $3.00 | 262,144 |
| Novita | novita | Not stated | $0.42 | $3.00 | 1,000,000 |
| Alibaba | alibaba | Not stated | $0.425 | $2.55 | 1,000,000 |
| Nebius | nebius/fp4 | fp4 | $0.45 | $3.00 | 1,000,000 |
| Cloudflare | cloudflare | Not stated | $0.45 | $3.20 | 262,144 |
| PrimeIntellect | primeintellect/fp8 | fp8 | $0.45 | $3.20 | 262,144 |
| Venice | venice/fp8 | fp8 | $0.45 | $3.20 | 262,144 |
| Cerebras | cerebras/fp16 | fp16 | $0.99 | $1.49 | 65,536 |
Sorted by input price, then output price. Precision is the quantization field of the endpoints list.
Measured by us our own short test request: about 380 tokens in, a short JSON answer out · medians from answers on two dates or more
United States
| Route | First answer token | Output speed | Cost per request | Hidden reasoning |
|---|---|---|---|---|
| Hosted by Cerebrasreasoning: off115 answers on 23 dates · 100 % answered | 139 msrange 0.10–16.85 s · slow 10 %: 259 ms Observed | 610 tokens/srange 199–627 | $0.00023range $0.000229–$0.00023 Calculated | 0for 64 visible |
| Hosted by Groqreasoning: off115 answers on 23 dates · 100 % answered | 195 msrange 0.10–1.09 s · slow 10 %: 391 ms Observed | 494 tokens/srange 210–528 · slowest 10 %: 428 tokens/s | $0.000549range $0.000546–$0.000553 Calculated | Hidden by providerthe API does not report it |
United Kingdom
| Route | First answer token | Output speed | Cost per request | Hidden reasoning |
|---|---|---|---|---|
| Hosted by Cerebrasreasoning: off105 answers on 21 dates · 100 % answered | 202 msrange 0.16–2.12 s · slow 10 %: 391 ms Observed | 546 tokens/srange 228–619 | $0.00023range $0.000229–$0.00023 Calculated | 0for 64 visible |
| Hosted by Groqreasoning: off105 answers on 21 dates · 100 % answered | 99 msrange 0.05–1.31 s · slow 10 %: 303 ms Observed | 493 tokens/srange 216–531 · slowest 10 %: 426 tokens/s | $0.000549range $0.000546–$0.000553 Calculated | Hidden by providerthe API does not report it |
Germany
| Route | First answer token | Output speed | Cost per request | Hidden reasoning |
|---|---|---|---|---|
| Hosted by Cerebrasreasoning: off105 answers on 21 dates · 100 % answered | 201 msrange 0.16–2.18 s · slow 10 %: 364 ms Observed | 571 tokens/srange 451–599 | $0.00023range $0.000229–$0.00023 Calculated | 0for 64 visible |
| Hosted by Groqreasoning: off103 answers on 21 dates · 98 % answered | 97 msrange 0.07–1.21 s · slow 10 %: 268 ms Observed | 494 tokens/srange 396–528 · slowest 10 %: 449 tokens/s | $0.000549range $0.000546–$0.000553 Calculated | Hidden by providerthe API does not report it |
Canada
| Route | First answer token | Output speed | Cost per request | Hidden reasoning |
|---|---|---|---|---|
| Hosted by Cerebrasreasoning: off105 answers on 21 dates · 100 % answered | 128 msrange 0.09–2.31 s · slow 10 %: 492 ms Observed | 566 tokens/srange 287–606 | $0.00023range $0.000229–$0.00023 Calculated | 0for 64 visible |
| Hosted by Groqreasoning: off104 answers on 21 dates · 99 % answered | 187 msrange 0.06–2.58 s · slow 10 %: 559 ms Observed | 489 tokens/srange 186–559 · slowest 10 %: 427 tokens/s | $0.000549range $0.000546–$0.000553 Calculated | Hidden by providerthe API does not report it |
Australia
| Route | First answer token | Output speed | Cost per request | Hidden reasoning |
|---|---|---|---|---|
| Hosted by Cerebrasreasoning: off105 answers on 21 dates · 100 % answered | 294 msrange 0.26–1.87 s · slow 10 %: 814 ms Observed | 496 tokens/srange 179–626 | $0.00023range $0.000229–$0.00023 Calculated | 0for 64 visible |
| Hosted by Groqreasoning: off105 answers on 21 dates · 100 % answered | 63 msrange 0.06–5.07 s · slow 10 %: 231 ms Observed | 506 tokens/srange 212–524 · slowest 10 %: 492 tokens/s | $0.000549range $0.000546–$0.000553 Calculated | Hidden by providerthe API does not report it |
- Hidden reasoning tokens are billed as output, at the output price, even though they are not part of the answer.
How the test works and what each number means: methodology. The figures are also in our open data.
Quality signals public test results, each with its source and date
| Signal | Result | Source |
|---|---|---|
| Instruction followingIFEval, 50 prompts · run by Groq · reasoning: off | 94 % Observedof 47 prompts with every instruction followed | Our run of a fixed IFEval subset, checked by code94 % on 22 Oct 2026 · 92 % on 17 Oct 2026 · instruction level 95 % |
Figures the provider publishes (6)
| Benchmark | Provider's figure | Where the provider says it |
|---|---|---|
| Terminal-Bench 2.1 | 73.0 Stated by the provider | Results table in Qwen/Qwen3.8-27B · Hugging Faceread 17 Oct 2026 |
| SWE-bench Pro | 61.7 Stated by the provider | Results table in Qwen/Qwen3.8-27B · Hugging Faceread 17 Oct 2026 |
| DeepSWE 1.1 | 42.2 Stated by the provider | Results table in Qwen/Qwen3.8-27B · Hugging Faceread 17 Oct 2026 |
| GPQA Diamond | 89.2 Stated by the provider | Results table in Qwen/Qwen3.8-27B · Hugging Faceread 17 Oct 2026 |
| Humanity's Last Exam (HLE) | 30.8 Stated by the provider | Results table in Qwen/Qwen3.8-27B · Hugging Faceread 17 Oct 2026 |
| LiveCodeBench v6 | 90.3 Stated by the provider | Results table in Qwen/Qwen3.8-27B · Hugging Faceread 17 Oct 2026 |
- Your own task: quality on it is not measured here. These are public test sets; a short trial on a few of your own examples is the check that fits your work.
- Other public leaderboards, without figures copied here: human preference votes on open-ended answers: Arena (LMArena) text leaderboard; long tasks in coding, data analysis and research: LiveBench.
What each signal covers and how our run is scored: methodology.
How we got this number
- What we read: the published price on openrouter.ai/api/v1/models
- Checked: ; tracked since 16 Oct 2026
- Shown in: the United States
- Checks so far: 25 checks since 16 Oct 2026
- What changed: the price last changed on ; every change is listed under History
- Page updated:
Other measured facts
| Question | Answer | Checked | Status |
|---|---|---|---|
| How much does Qwen Qwen3.8 27B cost per 1M cache-read tokens (cache hits) on OpenRouter? | $0.04 per 1M tokens | Published | |
| How much does Qwen Qwen3.8 27B cost per 1M output tokens on OpenRouter? | $1.88 per 1M tokens | Published | |
| How much does Qwen3.8 27B on Groq cost per 1M input tokens directly from Groq (official API price)? | $0.80 per 1M tokens | Published | |
| How much does Qwen3.8 27B on Groq cost per 1M output tokens directly from Groq (official API price)? | $4.00 per 1M tokens | Published | |
| What is the context window of Qwen Qwen3.8 27B on OpenRouter, in tokens? | 1,024,000 tokens | Published | |
| What is the maximum output length of Qwen Qwen3.8 27B on OpenRouter, in tokens? | 235,929 tokens | Published |
History
- — 1,000,000 tokens → 1,024,000 tokens (What is the context window of Qwen Qwen3.8 27B on OpenRouter, in tokens?) Confirmed by the check on 4 Nov 2026. What this means for you: The figure you would rely on has moved; check the evidence before relying on the old one.
- — $0.0039 per 1M tokens → $0.04 per 1M tokens (How much does Qwen Qwen3.8 27B cost per 1M cache-read tokens (cache hits) on OpenRouter?) Confirmed by the check on 1 Nov 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $0.04 per 1M tokens → $0.15 per 1M tokens Confirmed by the check on 1 Nov 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $1.25 per 1M tokens → $1.88 per 1M tokens (How much does Qwen Qwen3.8 27B cost per 1M output tokens on OpenRouter?) Confirmed by the check on 1 Nov 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $2.50 per 1M tokens → $1.25 per 1M tokens (How much does Qwen Qwen3.8 27B cost per 1M output tokens on OpenRouter?) Confirmed by the check on 29 Oct 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $0.20 per 1M tokens → $0.04 per 1M tokens Confirmed by the check on 29 Oct 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $0.04 per 1M tokens → $0.20 per 1M tokens Confirmed by the check on 25 Oct 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $1.35 per 1M tokens → $2.50 per 1M tokens (How much does Qwen Qwen3.8 27B cost per 1M output tokens on OpenRouter?) Confirmed by the check on 25 Oct 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $2.20 per 1M tokens → $1.35 per 1M tokens (How much does Qwen Qwen3.8 27B cost per 1M output tokens on OpenRouter?) Confirmed by the check on 22 Oct 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — $0.24 per 1M tokens → $0.04 per 1M tokens Confirmed by the check on 22 Oct 2026. What this means for you: If you buy or renew this plan now, you pay the new price; the old price is shown for comparison.
- — First check: $0.24 per 1M tokens
- Short-lived or unconfirmed moves are not shown as changes.
The source record lists $0.00000015 per token ($0.15 per 1M tokens), fetched on 9 Nov 2026.
The source record, as published
"prompt":"0.00000015"
Latest briefings each price change recalculated as a task cost
- DeepSeek V4 Pro 0813: price rise input $0.115 → $0.111 per 1M tokens (-3 %), output $4.00 → $4.44 per 1M tokens (+11 %)
- DeepSeek V4.1 Flash 1017: price cut input $0.300 → $0.0060 per 1M tokens (-98 %), output $1.20 → $0.500 per 1M tokens (-58 %)
- MoonshotAI Kimi K2.5: price rise input $0.500 → $0.532 per 1M tokens (+6 %), output $2.50 → $3.33 per 1M tokens (+33 %)
Record history — 147 records for 7 questions
- Last record
- Shown in
- the United States
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | $0.15 per 1M tokens | Published |
| 8 Nov 2026 | $0.15 per 1M tokens | Published |
| 7 Nov 2026 | $0.15 per 1M tokens | Published |
| 6 Nov 2026 | $0.15 per 1M tokens | Published |
| 5 Nov 2026 | $0.15 per 1M tokens | Published |
| 4 Nov 2026 | $0.15 per 1M tokens | Published |
| 3 Nov 2026 | $0.15 per 1M tokens | Published |
| 2 Nov 2026 | $0.15 per 1M tokens | Published |
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | $0.04 per 1M tokens | Published |
| 8 Nov 2026 | $0.04 per 1M tokens | Published |
| 7 Nov 2026 | $0.04 per 1M tokens | Published |
| 6 Nov 2026 | $0.04 per 1M tokens | Published |
| 5 Nov 2026 | $0.04 per 1M tokens | Published |
| 4 Nov 2026 | $0.04 per 1M tokens | Published |
| 3 Nov 2026 | $0.04 per 1M tokens | Published |
| 2 Nov 2026 | $0.04 per 1M tokens | Published |
| 1 Nov 2026 | $0.04 per 1M tokens | Published |
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | $1.88 per 1M tokens | Published |
| 8 Nov 2026 | $1.88 per 1M tokens | Published |
| 7 Nov 2026 | $1.88 per 1M tokens | Published |
| 6 Nov 2026 | $1.88 per 1M tokens | Published |
| 5 Nov 2026 | $1.88 per 1M tokens | Published |
| 4 Nov 2026 | $1.88 per 1M tokens | Published |
| 3 Nov 2026 | $1.88 per 1M tokens | Published |
| 2 Nov 2026 | $1.88 per 1M tokens | Published |
| 1 Nov 2026 | $1.88 per 1M tokens | Published |
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | $0.80 per 1M tokens | Published |
| 8 Nov 2026 | $0.80 per 1M tokens | Published |
| 7 Nov 2026 | $0.80 per 1M tokens | Published |
| 6 Nov 2026 | $0.80 per 1M tokens | Published |
| 5 Nov 2026 | $0.80 per 1M tokens | Published |
| 4 Nov 2026 | $0.80 per 1M tokens | Published |
| 3 Nov 2026 | $0.80 per 1M tokens | Published |
| 2 Nov 2026 | $0.80 per 1M tokens | Published |
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | $4.00 per 1M tokens | Published |
| 8 Nov 2026 | $4.00 per 1M tokens | Published |
| 7 Nov 2026 | $4.00 per 1M tokens | Published |
| 6 Nov 2026 | $4.00 per 1M tokens | Published |
| 5 Nov 2026 | $4.00 per 1M tokens | Published |
| 4 Nov 2026 | $4.00 per 1M tokens | Published |
| 3 Nov 2026 | $4.00 per 1M tokens | Published |
| 2 Nov 2026 | $4.00 per 1M tokens | Published |
| 1 Nov 2026 | $4.00 per 1M tokens | Published |
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | 1,024,000 tokens | Published |
| 8 Nov 2026 | 1,024,000 tokens | Published |
| 7 Nov 2026 | 1,024,000 tokens | Published |
| 6 Nov 2026 | 1,024,000 tokens | Published |
| 5 Nov 2026 | 1,024,000 tokens | Published |
| 4 Nov 2026 | 1,024,000 tokens | Published |
| 3 Nov 2026 | 1,024,000 tokens | Published |
| 2 Nov 2026 | 1,000,000 tokens | Published |
| Time | Value | Status |
|---|---|---|
| 9 Nov 2026 | 235,929 tokens | Published |
| 8 Nov 2026 | 235,929 tokens | Published |
| 7 Nov 2026 | 235,929 tokens | Published |
| 6 Nov 2026 | 235,929 tokens | Published |
| 5 Nov 2026 | 235,929 tokens | Published |
| 4 Nov 2026 | 235,929 tokens | Published |
| 3 Nov 2026 | 235,929 tokens | Published |
| 2 Nov 2026 | 235,929 tokens | Published |
| 1 Nov 2026 | 235,929 tokens | Published |
Showing the latest 60 of 147 records; the full history is in data.json and data.csv.