Qwen3.8 27B vs Qwen3.6 27B
Qwen3.8 27B scores higher on the intelligence index, 34 against 22. Both take the same machine to start: the cheapest here that runs either is the Framework Desktop, 32GB, at $1,269. On it, they run at much the same speed: 10 and 10 tok/s, both estimated from memory bandwidth. At 500k tokens a day the Framework Desktop, 32GB pays for itself in 22 years running Qwen3.6 27B, against 91 years running Qwen3.8 27B.
| Qwen3.8 27B | Qwen3.6 27B | |
|---|---|---|
| Intelligence index | 34 | 22 |
| Class | Sonnet-class | Haiku-class |
| Weights | 16 GB | 17 GB |
| Needs at 32k | 19 GB | 19 GB |
| Quantisation | Q4_K_M | Q4_K_M |
| Parameters | 27.8B | 27.8B |
| Max context | 256k | 256k |
| API price per 1M | $0.039 in / $1.25 out | $0.25 in / $2 out |
| Licence | Apache 2.0 | Apache 2.0 |
| Machines here that run it | 30 of 37 | 30 of 37 |
| Cheapest machine that runs it | Strix Halo Framework Desktop, 32GB $1,269 | Strix Halo Framework Desktop, 32GB $1,269 |
Run Qwen3.8 27B on the Framework Desktop, 32GB · or Qwen3.6 27B
Jump to: Side by side on the Framework Desktop, 32GB · How much use it takes to pay for the machine · Memory is not what separates them · The assumptions behind both columns
Side by side on the Framework Desktop, 32GB
The cheapest machine that runs either model is the same one, so this is the pair doing the same work on the same hardware: Strix Halo Framework Desktop, 32GB, at $1,269.
| Qwen3.8 27B | Qwen3.6 27B | |
|---|---|---|
| Speed at 32k | 10 tok/s estimated | 10 tok/s estimated |
| Pay-back on this machine | Pays back in 91 years | Pays back in 22 years |
| API cost per month | $1.75 | $5.47 |
Run Qwen3.8 27B on the Framework Desktop, 32GB · or Qwen3.6 27B
How much use it takes to pay for the machine
Everything above is at 500k tokens a day. Pay-back moves with how much you actually run, so here are both models at the five levels of use the calculator names, on the Framework Desktop, 32GB. Qwen3.6 27B pays for it sooner at every level of use, so which of them to run does not turn on how hard you work it.
| A day's use | Qwen3.8 27B | Qwen3.6 27B |
|---|---|---|
| 50ka few chats a day | 907 years | 217 years |
| 200klight assistant use | 227 years | 54 years |
| 1Ma moderate coding-assistant day | 45 years | 11 years |
| 4Mheavy coding with an agent | 11 years | 2.7 years |
| 20Magents running most of the day | 3.2 yearsits ceiling | 9.3 monthsits ceiling |
The Framework Desktop, 32GB cannot generate 20M tokens a day on either model: at most 14.3M on Qwen3.8 27B and 14M on Qwen3.6 27B. Both figures on that row are for the most it can do.
Run Qwen3.8 27B at 20M tokens a day · or Qwen3.6 27B
Memory is not what separates them
Both need 19 GB of memory at 32k of context. Every machine priced here that runs one runs the other, at 32k of context. So the choice between them is what each is good at and what the same work costs on an API, not what you have to buy to hold it.
Qwen3.8 27B is also head to head with DeepSeek V4-Flash above it on the leaderboard, Beam 1.0 501B-A23B below it and Qwen3 32B, the last-generation Qwen nearest it in size. One more model needs much the same memory: Arete-27B. Qwen3.6 27B is also head to head with MiniMax M2.7 above it on the leaderboard and Nemotron 3.5 Bolt 74B-A9B below it. One more model needs much the same memory: Arete-27B.
The assumptions behind both columns
Both columns use the same usage: 500k tokens a day at 15:1 input to output, 32k of context, $0.17 per kWh, and today's API prices held flat. Speeds marked estimated are worked out from memory bandwidth rather than measured, and pay-back scales with them. Where nobody rents an open model by the token, its API prices are the nearest hosted model's, named beside them. Machines are the 37 here with a published price that are still sold. Change any of it in the calculator.
More head to head: every machine that runs Qwen3.8 27B · every machine that runs Qwen3.6 27B · every other match-up · both against the frontier · the quickest pay-back at each level of use