Qwen3.5 4B vs gpt-oss-120b
Qwen3.5 4B scores higher on the intelligence index, 13 against 12. The cheapest machine here that runs gpt-oss-120b is the Framework Desktop, 128GB, at $3,449. Qwen3.5 4B runs on the Mac mini M6, 16GB at $899, $2,550 less. On the Framework Desktop, 128GB, the cheapest machine here that runs both, Qwen3.5 4B is about 1.4× quicker: 50 tok/s against 35, gpt-oss-120b's measured and the other estimated from memory bandwidth. At 500k tokens a day the Framework Desktop, 128GB pays for itself in 327 years running Qwen3.5 4B, against 720 years running gpt-oss-120b.
| Qwen3.5 4B | gpt-oss-120b | |
|---|---|---|
| Intelligence index | 13 | 12 |
| Class | Below every hosted tier | Below every hosted tier |
| Weights | 2.7 GB | 63 GB |
| Needs at 32k | 3.8 GB | 65 GB |
| Quantisation | Q4_K_M | MXFP4 |
| Parameters | 4.7B | 116.8B (5.1B active) |
| Max context | 256k | 128k |
| API price per 1M | $0.06 in / $0.15 outpriced as Granite 4.2 8B | $0.03 in / $0.15 out |
| Licence | Apache 2.0 | Apache 2.0 |
| Machines here that run it | 37 of 37 | 13 of 37 |
| Cheapest machine that runs it | Mac mini M6, 16GB $899 | Strix Halo Framework Desktop, 128GB $3,449 |
Run Qwen3.5 4B on the Mac mini M6, 16GB · or gpt-oss-120b on the Framework Desktop, 128GB
Jump to: Side by side on the Framework Desktop, 128GB · How much use it takes to pay for the machine · Machines that run one and not the other · The assumptions behind both columns
Side by side on the Framework Desktop, 128GB
The table above gives each model the cheapest machine that runs it, and those are two different machines, so nothing in it is a like-for-like race. The Strix Halo Framework Desktop, 128GB is the cheapest machine here that runs both, so this is the pair doing the same work on the same hardware.
| Qwen3.5 4B | gpt-oss-120b | |
|---|---|---|
| Speed at 32k | 50 tok/s estimated | 35 tok/s measured |
| Pay-back on this machine | Pays back in 327 years | Pays back in 720 years |
| API cost per month | $1.00priced as Granite 4.2 8B | $0.57 |
Run Qwen3.5 4B on the Framework Desktop, 128GB · or gpt-oss-120b
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, 128GB. Qwen3.5 4B 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 | Qwen3.5 4B | gpt-oss-120b |
|---|---|---|
| 50ka few chats a day | 3,266 years | 7,203 years |
| 200klight assistant use | 816 years | 1,801 years |
| 1Ma moderate coding-assistant day | 163 years | 360 years |
| 4Mheavy coding with an agent | 41 years | 90 years |
| 20Magents running most of the day | 8.2 years | 18 years |
Run Qwen3.5 4B at 20M tokens a day · or gpt-oss-120b
Machines that run one and not the other
Qwen3.5 4B needs 3.8 GB of memory at 32k of context and gpt-oss-120b needs 65 GB. That puts Qwen3.5 4B on 24 of the 37 machines priced here that gpt-oss-120b does not, starting at $899.
| Machine | Price | Memory | Speed on Qwen3.5 4B | Pay-back |
|---|---|---|---|---|
| Mac mini M6, 16GB | $899 | 16 GB | 30 tok/s estimated | Pays back in 83 years |
| Mac mini M6, 24GB | $1,099 | 24 GB | 33 tok/s estimated | Pays back in 100 years |
| Strix Halo Framework Desktop, 32GB | $1,269 | 32 GB | 50 tok/s estimated | Pays back in 120 years |
| Mac mini M6, 32GB | $1,299 | 32 GB | 33 tok/s estimated | Pays back in 119 years |
| MacBook Air M5 (13-inch), 16GB | $1,299 | 16 GB | 30 tok/s estimated | Pays back in 120 years |
| AMD Radeon AI PRO R9700, 32GB | $1,299card only | 32 GB | 146 tok/s estimated | Pays back in 119 years |
18 more, on the Qwen3.5 4B page.
Qwen3.5 4B is also head to head with MiniCPM5 2B above it on the leaderboard. gpt-oss-120b is also head to head with Ling 3.0 tiny below it on the leaderboard. One more model needs much the same memory: Llama 4 Scout 17B-16E.
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 Qwen3.5 4B · every machine that runs gpt-oss-120b · every other match-up · both against the frontier · the quickest pay-back at each level of use