meta-llama/Llama-3.1-8B-Instruct| Provider | Input, $ per 1M | Output, $ per 1M | In our data since |
|---|---|---|---|
| novita-ai | $0.02 | $0.05 | 31 Jul 2026 |
| Abacus.AI | $0.02 | $0.05 | 1 Aug 2026 |
| Helicone | $0.02 | $0.05 | 31 Jul 2026 |
| Kilo Code | $0.02 | $0.05 | 31 Jul 2026 |
NanoGPT
Meta-Llama-3-1-8B-Instruct-FP8
|
$0.02 | $0.03 | 1 Aug 2026 |
| Nebius Token Factory | $0.02 | $0.06 | 1 Aug 2026 |
| Novita AI | $0.02 | $0.05 | 31 Jul 2026 |
| inference | $0.025 | $0.025 | 31 Jul 2026 |
| DeepInfra | $0.03 | $0.05 | 1 Aug 2026 |
| Nscale | $0.03 | $0.03 | 31 Jul 2026 |
| regolo-ai | $0.05 | $0.25 | 31 Jul 2026 |
| OpenRouter | $0.05 | $0.08 | 31 Jul 2026 |
NanoGPT
meta-llama/llama-3.1-8b-instruct
|
$0.054 | $0.054 | 31 Jul 2026 |
| friendliai | $0.1 | $0.1 | 1 Aug 2026 |
| OVHCloud AI Endpoints | $0.1 | $0.1 | 31 Jul 2026 |
| SambaNova | $0.1 | $0.2 | 1 Aug 2026 |
Cloudflare AI Gateway
workers-ai/@cf/meta/llama-3.1-8b-instruct-awq
|
$0.12 | $0.27 | 31 Jul 2026 |
| Hyperbolic | $0.12 | $0.3 | 1 Aug 2026 |
| neon | $0.15 | $0.45 | 1 Aug 2026 |
Cloudflare AI Gateway
workers-ai/@cf/meta/llama-3.1-8b-instruct-fp8
|
$0.15 | $0.29 | 31 Jul 2026 |
| Databricks | $0.15 | $0.45 | 1 Aug 2026 |
| cloudflare-workers-ai | $0.152 | $0.287 | 31 Jul 2026 |
| Cloudflare Workers AI | $0.152 | $0.287 | 31 Jul 2026 |
| Perplexity | $0.2 | $0.2 | 31 Jul 2026 |
| Pioneer | $0.2 | $0.2 | 31 Jul 2026 |
| Amazon Bedrock | $0.22 | $0.22 | 2 Aug 2026 |
| Weights & Biases | $0.22 | $0.22 | 31 Jul 2026 |
Cloudflare AI Gateway
workers-ai/@cf/meta/llama-3.1-8b-instruct
|
$0.28 | $0.83 | 31 Jul 2026 |
| Microsoft Foundry | $0.3 | $0.61 | 1 Aug 2026 |
| OCI Generative AI | $0.72 | $0.72 | 31 Jul 2026 |
Only the base tier and only per-token prices are shown. Batch, discounted and cached rates, as well as prices per image or per second of video, do not go into this table: they cannot stand in the same column as a price per million tokens.
A single provider appears several times if it sells this model under different identifiers and at different prices — for example, at different weight precision. The identifier is shown next to the name. Identical offers that arrived from two sources under different spellings are merged into one row.
The date in the last column is the day this price first entered our collection. The price may well be older: before that day we simply were not recording it. It has not changed since — otherwise a new row with a new date would stand in its place.
| Task set | Result | Run conditions | Measured by |
|---|---|---|---|
| Arena Score in English | 1,217.68 1,213–1,223 | — | — |
| Arena Score, programming | 1,195.29 1,188–1,203 | — | — |
| Arena Score, overall | 1,186.74 1,183–1,191 | — | — |
| Arena Score, long queries | 1,181.83 1,174–1,190 | — | — |
| Arena Score, mathematics | 1,179.26 1,171–1,187 | — | — |
| Arena Score in French | 1,176.17 1,147–1,206 | — | — |
| Arena Score, hard prompts | 1,174.99 1,169–1,181 | — | — |
| Arena Score, multi-turn dialogue | 1,172.04 1,164–1,180 | — | — |
| Arena Score in Spanish | 1,168.43 1,141–1,196 | — | — |
| Arena Score, instruction following | 1,159.42 1,154–1,165 | — | — |
| Arena Score in Russian | 1,157.4 1,149–1,166 | — | — |
| Arena Score, creative writing | 1,154.71 1,146–1,163 | — | — |
| Arena Score, expert questions | 1,143.55 1,131–1,156 | — | — |
| MATH, difficulty level five | 22.88 % | · with a tuned harness | Epoch evaluations |
| BALROG — game environments | 15.1 % | — | Balrog Leaderboard |
| Mock AIME 2024–2025 — olympiad problems | 2.4 % | · with a tuned harness | Epoch evaluations |
| WeirdML — unusual machine learning tasks | 1.73 % | — | WeirdML Leaderboard |
| GPQA Diamond — graduate-level questions | 1.26 % | · with a tuned harness | Epoch evaluations |
| CritPt — physics problems | 0 % | — | — |