Sunk Cost sunkcost.ai Data checked 2026-09-03

Llama 3.1 8B Instruct vs Ministral 3 8B

Llama 3.1 8B Instruct scores higher on the intelligence index, 7 against 5. Both take the same machine to start: the cheapest here that runs either is the Mac mini M6, 16GB, at $899. On it, they run at much the same speed: 12 and 12 tok/s, both estimated from memory bandwidth. At 500k tokens a day the Mac mini M6, 16GB pays for itself in 37 years running Ministral 3 8B, against 842 years running Llama 3.1 8B Instruct.

Llama 3.1 8B InstructMinistral 3 8B
Intelligence index75
ClassBelow every hosted tierBelow every hosted tier
Weights4.9 GB5.2 GB
Needs at 32k9.2 GB9.8 GB
QuantisationQ4_K_MQ4_K_M
Parameters8B8.9B
Max context128k256k
API price per 1M$0.02 in / $0.04 out$0.15 in / $0.15 out
LicenceLlama 3.1 Community LicenseApache 2.0
Machines here that run it37 of 3737 of 37
Cheapest machine that runs itMac mini M6, 16GB $899Mac mini M6, 16GB $899
Summarising good not rated
Translation usable not rated
Everyday coding usable not rated
Reasoning & maths don’t not rated
Agentic work don’t not rated

Run Llama 3.1 8B Instruct on the Mac mini M6, 16GB · or Ministral 3 8B

Ratings are coarse on purpose: they say what a model is usable for, not where it places to the decimal.

Side by side on the Mac mini M6, 16GB

The cheapest machine that runs either model is the same one, so this is the pair doing the same work on the same hardware: Mac mini M6, 16GB, at $899.

Llama 3.1 8B InstructMinistral 3 8B
Speed at 32k12 tok/s estimated12 tok/s estimated
Pay-back on this machinePays back in 842 yearsPays back in 37 years
API cost per month$0.32$2.28

Run Llama 3.1 8B Instruct on the Mac mini M6, 16GB · or Ministral 3 8B

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 Mac mini M6, 16GB. Ministral 3 8B 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 useLlama 3.1 8B InstructMinistral 3 8B
50ka few chats a day8,423 years368 years
200klight assistant use2,106 years92 years
1Ma moderate coding-assistant day421 years18 years
4Mheavy coding with an agent105 years4.6 years
20Magents running most of the day24 yearsits ceiling14 monthsits ceiling

The Mac mini M6, 16GB cannot generate 20M tokens a day on either model: at most 17.2M on Llama 3.1 8B Instruct and 16.2M on Ministral 3 8B. Both figures on that row are for the most it can do.

Run Llama 3.1 8B Instruct at 20M tokens a day · or Ministral 3 8B

Memory is not what separates them

Llama 3.1 8B Instruct needs 9.2 GB of memory at 32k of context and Ministral 3 8B needs 9.8 GB. 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, how fast it runs and what the same work costs on an API, not what you have to buy to hold it.

Llama 3.1 8B Instruct is also head to head with Devstral Small 2 24B above it on the leaderboard and Qwen3 32B below it. Ministral 3 8B is also head to head with Gemma 3 27B it above it on the leaderboard and Gemma 3 12B it below it.

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 Llama 3.1 8B Instruct · every machine that runs Ministral 3 8B · every other match-up · both against the frontier · the quickest pay-back at each level of use