Head to head
IB Granite 4.0 Microvs ME Llama 3.1 8B InstructFor light moderate heavy tasks, about a month.
Verdict · measured basis
Granite 4.0 Micro costs $0.0000 per solved task against
Llama 3.1 8B Instruct at $0.0000 — ▼ 1.0x cheaper .
Over 200 tasks that is $0 a month.
On output token price alone Llama 3.1 8B Instruct looks 1.4x cheaper. The ranking reverses once you divide by pass rate — the cheaper tokens do not win the task.
Copy link Save scenario Export compare table Planned
Verified pricing, pass rate and cost per solved task for Granite 4.0 Micro and Llama 3.1 8B Instruct Figure Granite 4.0 Micro Llama 3.1 8B Instruct Input $/Mtok $0.017 $0.05 Output $/Mtok $0.112 $0.08 Cached input $/Mtok $0.017 $0.025 Pass rate 42% 42% Cost per solved task $0.0000 measured $0.0000 measured Cost per month $0.01 $0.01
Cost per month against tasks per month · drag the marker
Monthly cost against tasks per month Log-log lines for Granite 4.0 Micro (measured, $0.0000 per solved task), Llama 3.1 8B Instruct (measured, $0.0000 per solved task) from 10 to 100,000 tasks a month. Marker at 200 tasks. 10 100 1k 10k 100k $0.001 $0.01 $0.1 $1 TASKS / MONTH (log) → 200 tasks Granite 4.0 Micro · MEASURED · $0.0000 per solved task $0.01 Granite 4.0 Micro MEASURED Llama 3.1 8B Instruct · MEASURED · $0.0000 per solved task $0.01 Llama 3.1 8B Instruct MEASURED Monthly cost against tasks per month Log-log lines for Granite 4.0 Micro (measured, $0.0000 per solved task), Llama 3.1 8B Instruct (measured, $0.0000 per solved task) from 10 to 100,000 tasks a month. Marker at 200 tasks. 10 100 1k 10k 100k $0.001 $0.01 $0.1 $1 TASKS / MONTH (log) → 200 tasks Granite 4.0 Micro · MEASURED · $0.0000 per solved task $0.01 Llama 3.1 8B Instruct · MEASURED · $0.0000 per solved task $0.01 — Granite 4.0 Micro · MEASURED — Llama 3.1 8B Instruct · MEASURED Solid measured · dashed modeled · dotted stale. The gap is the verdict.
Solvency Bench (solvency.dev) — first-party measurement · verified 2026-08-26 · prices from each provider's pricing page; measured rows ignore the tier · Methodology · All models at these settings