Qwen3.8 27B API Pricing 2026: Cost and Cheapest Providers

Updated 20 October 2026. Qwen3.8 27B costs $0.05 per million input tokens and $0.28 output on SayGM, or about $9.50 per 100 million tokens at an 80/20 mix; the cheapest major host is DeepInfra at about $50, on a 25% promotional rate.

Buy direct from Alibaba instead if you want day-one access to new checkpoints; for the full 1M-token context, Alibaba, Novita and Phala are the three hosts that serve it. If you need the prompt sealed in hardware, Qwen3.8 27B has no sealed route on SayGM yet, so use Qwen3.6 27B (TEE) at $0.29 and $1.90.

Qwen3.8 27B API pricing, checked 20 October 2026

Table of Contents

• The cheapest Qwen3.8 27B provider
• Qwen3.8 27B providers compared
• Pricing for the rest of the Qwen family
• Qwen3.8 27B and confidential inference
• Qwen3.8 27B hosting: running it yourself
• Calling Qwen3.8 27B on SayGM
• Qwen3.8 27B pricing FAQ
• The bottom line
• About SayGM

Pricing source: SayGM from the Qwen3.8 27B model page; everyone else from OpenRouter's provider list. Twenty providers serve the model; this table keeps the major hosts. Output is where the bill is: at 80/20, output is 60% or more of the cost on every route. DeepInfra's undiscounted rate is $0.20 and $2.50, or $66 per 100M.

Alibaba released Qwen3.8 27B on 14 August 2026 under Apache 2.0, which permits commercial use with no revenue share. That licence is why 20 providers serve it and why the market price sits so far below retail.

It is a 27-billion-parameter dense model, not a mixture of experts, with a 262K native context extending to 1M through YaRN, and it takes text, images and video as input. On Alibaba's own evaluations, reported in the release coverage, it scores 61.7% on SWE-bench Pro against 53.4% for Claude Opus 4.6 Max, and 42.2% on DeepSWE 1.1 against 13.3% for Qwen3.6-27B. Those figures have not been independently reproduced.

The published retail rate is $0.50 input and $3.00 output per million tokens, though most providers charge well below.

Qwen3.8 27B providers compared

The providers most teams will consider, and what each one is good for:

SayGM

$0.05 in and $0.28 out, with cached reads at $0.01, which works out at about $9.50 per 100 million tokens, around a fifth of DeepInfra's promotional rate. That is the standing rate in our published catalog, not a promotional tier: providers compete to serve requests and the rate falls out of that. There is no subscription and no top-up fee. What SayGM does not offer on this model: a sealed route (qwen3.8-27b is standard, see below), dedicated deployments, fine-tuning, or a published precision. If you need the cheapest way to run a Qwen agent loop at volume, this is it.

Alibaba Cloud (direct)

Going to the source buys two real things: the full 1M context window, which Novita and Phala also serve, and first access to new checkpoints. Uptime was 98.95% over the 24 hours to 20 October on OpenRouter's tracking. The cost is the cost, 1.7 times DeepInfra and about nine times SayGM, and requests are processed by Alibaba.

OpenRouter

Where most developers first meet a new open-weight model, and for Qwen3.8 27B it puts 20 providers behind one endpoint. Inference is passed through at each provider's rate; the cost is a fee on credit top-ups, 5.5% by card and 5% by crypto. What you get is routing: sort modes for price, speed or throughput, automatic failover when a provider errors, per-provider uptime and precision data, and data-policy filters so you can exclude providers that train on prompts. The endpoint data in the table above is OpenRouter's, and it is the best public source on this model. What a router structurally cannot do is seal the prompt: it sees every request in the clear, and its privacy is whatever the downstream provider's policy says. We set the two models out in SayGM vs OpenRouter.

DeepInfra

$0.15 and $1.875 on a 25% promotional rate, $50 per 100 million ($66 at the $0.20 and $2.50 list price), and the only provider serving Qwen3.8 27B at BF16. If you are comparing quality across cheap routes, this is the reference: Cerebras serves FP16 at nearly twice the price, and everyone else is at FP8 or lower, or has not said. Cache reads are $0.0375, which matters on repeated system prompts. Uptime was 100% over the 24 hours to 20 October on OpenRouter's tracking. It is roughly five times SayGM on a normal mix, and there is no confidential compute option, but for a team that wants unquantised weights and a mainstream vendor it is the straightforward pick.

Cloudflare Workers AI

$0.45 and $3.20, $100 per 100 million, the same as buying at Qwen's retail rate, with 100% uptime over the 24 hours to 20 October on OpenRouter's tracking. What you are buying is adjacency: if your application already runs on Workers, inference at the edge next to it removes a network hop and a vendor. That is a real architectural argument. It is not a price argument.

Pricing for the rest of the Qwen family

Qwen3.8 27B is the current open-weight release, but plenty of production traffic still runs against earlier versions, and Alibaba has shipped four closed 3.8 models since August. SayGM rates are from each model page; the models SayGM doesn't serve are priced from OpenRouter. All checked 20 October 2026.

The cheapest sealed Qwen is now Qwen3 32B (TEE) at $0.099 and $0.40, about 1.7 times the standard 3.8 27B route per 100M tokens. The four newer 3.8 models aren't on SayGM yet, so their rows are OpenRouter's prices. The full catalog has current rates for every route, and our pricing page covers billing.

What changed since September

This post first went up on 25 September 2026. Here's what has moved since.

• SayGM went from $0.05 and $0.30 to $0.05 and $0.28, which takes 100M tokens from $10 to about $9.50.
• DeepInfra is still $0.15 and $1.875, but OpenRouter now marks that as 25% off a $0.20 and $2.50 list price, so expect it to rise when the promotion ends.
• Alibaba, CoreWeave, Novita and Cloudflare have not moved.
• SayGM's sealed Qwen routes went up. Qwen3.6 27B (TEE) went from $0.147 and $0.982 to $0.29 and $1.90, and Qwen3.5 397B (TEE) from $0.155 and $1.034 to $0.43 and $2.85.
• New Qwen models since: Qwen3.8 Flash (26 August), Qwen3.8 Max (3 September), Qwen3.8 Omni Flash (21 September) and Qwen3.8 Max Prime (23 September). None is on SayGM; their prices are in the family table above.

Qwen3.8 27B hosting: running it yourself

The weights are Apache 2.0, so self-hosting is a real option in a way it is not for a frontier model. What it takes, per Yotta Labs' hardware breakdown:

Those figures are the weights alone. KV cache sits on top and grows with context length and concurrency, which is what usually pushes a 56GB model onto an 80GB card.

At 4-bit on a 4090 this is genuinely runnable on a desk, which makes it a reasonable choice for local development. For production, do the comparison directly: take your monthly token volume at about $9.50 per 100 million, then price an always-on 80GB instance at your cloud's hourly rate and add the ops time. Self-hosting wins when the instance is busy most of the day and you need the weights on your own hardware for other reasons.

Qwen3.8 27B and confidential inference

Private Inference is coming soon to qwen3.8-27b.

For a confidential Qwen, use one of the four sealed routes in the family table above; the cheapest is qwen3-32b-tee at $0.099 and $0.40. Phala also serves Qwen3.8 27B in an enclave, at $0.20 and $2.50.

Calling Qwen3.8 27B on SayGM

A base URL and a model ID:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.saygm.com/v1",
    api_key="YOUR_SAYGM_KEY",
)

response = client.chat.completions.create(
    model="qwen3.8-27b",
    messages=[{"role": "user", "content": "Extract the line items from this invoice."}],
)
print(response.choices[0].message.content)

The Qwen3.8 27B model page carries the live rate, throughput and success rate, and the quickstart covers keys.

Qwen3.8 27B pricing FAQ

How much does the Qwen 3.8 27B API cost?

As of 20 October 2026, the Qwen 3.8 27B API costs $0.05 per million input tokens and $0.28 output on SayGM, against a published retail rate of $0.50 and $3.00. The major hosts in our table charge $0.15 to $0.45 on input and $1.875 to $3.20 on output. At an 80/20 mix that's about $9.50 per 100 million tokens on SayGM, $50 on DeepInfra and $85 direct from Alibaba.

What is the cheapest Qwen3 API?

As of October 2026, the cheapest Qwen3.8 27B API is SayGM at about $9.50 per 100 million tokens. Among the major hosts, DeepInfra is next at $50 on a promotional rate. For the older Qwen3 models, Qwen3 32B (TEE) on SayGM is $0.099 and $0.40, and it's sealed.

What does Qwen 3.8 27B cost on OpenRouter?

On 20 October 2026, OpenRouter lists 20 providers for Qwen3.8 27B, from $0.035 to $0.99 on input and $1.25 to $3.20 on output. OpenRouter passes the provider's rate through and charges a 5.5% fee on card top-ups, so the real cost is the table price plus 5.5%.

Does Featherless serve Qwen 3.6 27B, and what does it cost?

As of October 2026, Featherless sells plans rather than a single per-token price. Chat is $25 a month for unlimited tokens at a 32K context, but only for human-typed chat, with no API or agent traffic allowed. API use goes on the Developer plan, from $50 a month in credits billed per token at each model's rate.

Is there a free Qwen3.8 27B API?

Yes, as of 20 October 2026: OpenRouter lists a free variant at qwen/qwen3.8-27b:free. Free endpoints on routers are rate-limited and tend to disappear when the sponsoring provider's budget does, so they suit experiments rather than a product you're shipping.

Is Qwen3.8 27B cheaper than Kimi, GLM or DeepSeek?

As of October 2026, yes, with one near-tie: at about $9.50 per 100 million tokens on SayGM, Qwen3.8 27B is cheaper than every Kimi and DeepSeek route we carry, and level with GLM 5.3 Flash at about $10. The trade is capability at the top end. See Kimi API pricing, GLM pricing, DeepSeek API pricing and the cheapest LLM API comparison.

Qwen3.8 27B or Qwen3.6 27B?

3.8, unless you have validated against 3.6 and cannot move. The jump on DeepSWE 1.1 is 13.3% to 42.2% on Alibaba's evals, and 3.8 adds image and video input that 3.6 does not have.

Can Qwen3.8 27B run on a single GPU?

Yes. 4-bit quantised it needs 14 to 16GB, which fits a 4090. At full BF16 precision it needs about 56GB plus KV cache, so an 80GB card.

About SayGM

SayGM is a drop-in inference gateway for teams who don't want to just take a company's word that their prompts are private. Route to a confidential model and your prompt is sealed inside hardware-verified execution, where not even SayGM can see it. That's not a policy, it's provable. Swap in your existing OpenAI, Anthropic, or Gemini code and you're covered in minutes, at transparent, published rates with no hidden markup.

Say gm to AI at saygm.com.

Website | Twitter | Discord | Blog | Medium | Docs