
Llama 3.3 70B Instruct
Model ranking
- Arena Score
- 1274.2912307868
- Arena reference rank
- 286
- Evaluation configuration
- Llama 3.3 70B Instruct
- Source published
- 2026-11-19
- Votes
- 54473
- Confidence lower
- 1270.8013561268
- Confidence upper
- 1277.7811054467
- Last successful check
- 2026-11-22T14:00:05.877Z
- Last value/rank change
- 2026-11-22T09:30:20.053Z
- Methodology
- Human preference — Text / overall; checked every six hours.
Source: Arena / LMArena ↗ · CC BY 4.0. Selection and catalog matching by XPU LIVE.
| Provider | Price | Provider model ID | Status | Context / max output | Source |
|---|---|---|---|---|---|
Cloudflare Workers AIllama-3.3-70b-instruct-fp8-fast | Standard Tariff not listedInput $0.293USD / 1M tokensOutput $2.253USD / 1M tokens Updated 2026-11-22 · 15:43 UTCCloudflare Workers AI · StandardInput: $0.293 Output: $2.253 Published Workers AI USD list tariffs. Free daily Neuron allowance is separate, not a zero unit price. Image tile/step components must be evaluated together; no flat image price is inferred. Checked 2026-11-22T15:43:15.988Z Price source ↗ | @cf/meta/llama-3.3-70b-instruct-fp8-fast | Catalog-listedConditionsPublic catalog listing; account eligibility and capacity can vary. Verified component tariffs are supplied by the existing pricing engine. | Context tokens 24000 | Official source ↗2026-10-20 |
Router catalog Tariff not listedInput $0.71USD / 1M tokensOutput $0.71USD / 1M tokens Updated 2026-11-22 · 15:43 UTCOpenRouter · Router catalogInput: $0.71 Output: $0.71 Cache Read: $0.71 OpenRouter route catalog price, not a direct-provider price. Routing, modality and endpoint conditions can change the actual bill. Checked 2026-11-22T15:43:15.988Z Price source ↗ | meta-llama/llama-3.3-70b-instruct | Catalog-listedConditionsCatalog-listed; account access and regional availability may apply. | Context tokens 131072Output tokens 16384 | Official source ↗2026-09-28 | |
| Not collectedAwaiting verified pricing | Llama-3.3-70B-Instruct | Catalog-listedConditionsCatalog-listed; account access and regional availability may apply. | Not supplied | Official source ↗2026-09-28 |
Llama 3.3 70B Instruct is Meta’s instruction-tuned text model. Cloudflare’s FP8 Fast identifier describes a hosted deployment of that model, so it shares the existing canonical page.
Llama 3.3 70B Instruct: suitable applications
Evaluate multilingual dialogue, summarization and assistants that need a substantial open-weight instruction model.
Capabilities and technical details
The developer publishes weights under the Llama license, allowing an implementation to compare hosted and compatible self-managed deployments.
Llama 3.3 70B Instruct accepts text and returns text. The recorded input reference is 131,072 tokens, with any lower Cloudflare cap recorded separately. The recorded terms are Llama 3.3 Community License; the linked source explains their scope.
Llama 3.3 70B Instruct limitations and selection
The Workers AI context limit is lower than the reference model’s context. FP8 precision and the Fast deployment label are not independent model families or measured performance guarantees.
Check the endpoint’s actual input budget and compare answer quality on precision-sensitive prompts before substituting it for another Llama 3.3 deployment.
Llama 3.3 70B Instruct API pricing
Llama 3.3 70B Instruct pricing is tied to the selected endpoint and its documented token components. Compare input, output and cached-input charges only where those components are actually quoted; an absent component is not an invented free rate.
Review the Cloudflare Llama 3.3 70B Instruct endpoint reference and the Meta model documentation. Compare the linked Cloudflare Workers AI provider profile with the xpu live model directory when selecting a deployment.
Specifications
- Catalog model ID
- meta-llama/llama-3.3-70b-instruct
- Developer model ID / revision
- Not verified in cited sources
- Input modalities
- textCloudflare Workers AI endpoint; route-specific capabilities ↗
- Output modalities
- textCloudflare Workers AI endpoint; route-specific capabilities ↗
- Context / input token limit
- 131,072Developer model card ↗
- Maximum output tokens
- 16,384OpenRouter catalog ↗
- Tool calling
- SupportedCloudflare Workers AI endpoint; route-specific capabilities ↗
- Structured output
- SupportedOpenRouter catalog ↗
- API parameters (source-specific)
- frequency_penalty, logit_bias, logprobs, max_tokens, min_p, presence_penalty, repetition_penalty, response_format, seed, stop, structured_outputs, temperature, tool_choice, tools, top_k, top_logprobs, top_pOpenRouter catalog ↗
- API parameter scope
- OpenRouter API; support varies by route
- Weight access
- Open weightsDeveloper model card / documentation ↗
- Weight license / access terms
- Llama 3.3 Community LicenseDeveloper weight license / hosted access terms ↗
- Total parameters (billions)
- 70Developer model card / documentation ↗
- Active parameters (billions)
- Not verified in cited sources
- Parameter-count details
- Developer-declared nominal model size; serialized tensor counts can differ. Active parameters, when supplied, are separate from total parameters.Developer model card ↗
- Download size (GB)
- Depends on checkpoint and precision; no verified file size
- Release date
- Not verified in cited sources
- Added to OpenRouter
- 2024-12-06OpenRouter catalog ↗
- Notes
- Existing model reference context retained; Cloudflare FP8 Fast endpoint is capped at 24,000 input tokens. FP8 is a deployment format, not another model. Weight and service terms are separate. Parameter sizes are developer-declared nominal labels; serialized tensor counts may differ.
Catalog entries link to their specification and availability sources. This is a curated selection, not every model each provider offers. Best prices are the lowest collected USD tariffs, not a total-cost estimate. Token offers are selected by input price, with output from the same tariff; workload mix may change the cheapest provider. Listed variants may require batch processing, off-peak hours, or specific media settings. Price coverage varies by provider; collected tariffs include their source and observation time. Arena user-preference scores are checked automatically every six hours against the latest published Text / overall dataset. Source: Arena / LMArena; CC BY 4.0. Selection and catalog matching by XPU LIVE. They apply to the shown evaluation configuration; listed prices may use different settings. Models without verified scores are not assumed to be weaker.
