Qwen3.8 27B vs Beam 1.0 501B-A23B
Qwen3.8 27B scores higher on the intelligence index, 34 against 30. None of the 37 machines here still sold at a published price runs Beam 1.0 501B-A23B at 32k of context. Qwen3.8 27B runs on the Framework Desktop, 32GB, from $1,269. At 500k tokens a day that machine pays for itself in 17 years running Qwen3.8 27B.
| Qwen3.8 27B | Beam 1.0 501B-A23B | |
|---|---|---|
| Intelligence index | 34 | 30 |
| Class | Sonnet-class | Sonnet-class |
| Weights | 16 GB | 307 GB |
| Needs at 32k | 19 GB | 308 GB |
| Quantisation | Q4_K_M | UD-Q4_K_XL |
| Parameters | 27.8B | 501.3B (23.4B active) |
| Max context | 256k | 1M |
| API price per 1M | $0.32 in / $2.5 out | $0.25 in / $1 out |
| Licence | Apache 2.0 | Apache 2.0 |
| Machines here that run it | 30 of 37 | 0 of 37 |
| Cheapest machine that runs it | Strix Halo Framework Desktop, 32GB $1,269 | none listed |
Run Qwen3.8 27B on the Framework Desktop, 32GB · or Beam 1.0 501B-A23B in the calculator
Machines that run one and not the other
Qwen3.8 27B needs 19 GB of memory at 32k of context and Beam 1.0 501B-A23B needs 308 GB. That puts Qwen3.8 27B on 30 of the 37 machines priced here that Beam 1.0 501B-A23B does not, starting at $1,269.
| Machine | Price | Memory | Speed on Qwen3.8 27B | Pay-back |
|---|---|---|---|---|
| Strix Halo Framework Desktop, 32GB | $1,269 | 32 GB | 10 tok/s estimated | Pays back in 17 years |
| Mac mini M6, 32GB | $1,299 | 32 GB | 6.9 tok/s estimated | Pays back in 17 years |
| AMD Radeon AI PRO R9700, 32GB | $1,299card only | 32 GB | 30 tok/s estimated | Pays back in 17 years |
| Strix Halo Corsair AI Workstation 300, 64GB | $1,700 | 64 GB | 10 tok/s estimated | Pays back in 22 years |
| Strix Halo Framework Desktop, 64GB | $1,959 | 64 GB | 10 tok/s estimated | Pays back in 26 years |
| NVIDIA GeForce RTX 5090, 32GB | $1,999card only | 32 GB | 66 tok/s estimated | Pays back in 25 years |
24 more, on the Qwen3.8 27B page.
Qwen3.8 27B is also head to head with DeepSeek V4-Flash above it on the leaderboard and Qwen3 32B, the last-generation Qwen nearest it in size. One more model needs much the same memory: Muse Glimmer 30B. Beam 1.0 501B-A23B is also head to head with Inkling Small below it on the leaderboard.
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 · what Beam 1.0 501B-A23B needs · every other match-up · both against the frontier · the quickest pay-back at each level of use