Meta
llama-3.3-70b-instruct
$0.125 in · $0.4 out per million tokens · 131,072 tokens of context · first token in 1,988 ms
Price, limits and measured speed
- Time to first token
- 1,988 ms
- our probe of 22 November 2026 · measured by us
- Output speed
- 6.9 tokens a second
- our probe of 22 November 2026 · measured by us
- Price of our test call
- $0.000053
- our probe of 22 November 2026, as the API stated it · measured by us
- Input price rank
- 10th lowest of 55 chat models
- the models on this site we could call · measured by us
- Output price rank
- 9th lowest of 55
- the models on this site we could call · measured by us
- First-token rank
- 44th fastest of 55 measured
- our probe of 22 November 2026 · measured by us
- Output speed rank
- 42nd fastest of 42 measured
- our probe of 22 November 2026 · measured by us
- Context rank
- 40th largest of 55
- the models on this site we could call · measured by us
Call it
curl
curl https://api.plungeai.com/v1/chat/completions \
-H "Authorization: Bearer $PLUNGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "meta-llama/llama-3.3-70b-instruct",
"messages": [{"role": "user", "content": "Hello"}]
}'What it costs for three jobs
The same three jobs on every model page, so two models can be compared.
| Job | Tokens in | Tokens out | Through One API | At Meta's list price |
|---|---|---|---|---|
| A chat turn | 1,000 | 300 | $0.000245 | $0.00135 |
| An answer over retrieved documents | 8,000 | 500 | $0.0012 | $0.00884 |
| A summary of a long document | 100,000 | 2,000 | $0.0133 | $0.106 |
Models near it
Each against llama-3.3-70b-instruct: its input price and the time to its first token.
| Model | Why it is here | Input price against it | First token against it | Input, $ per 1M | Output, $ per 1M | Context, tokens | First token | Tokens a second |
|---|---|---|---|---|---|---|---|---|
| deepinfra/meta-llama/Llama-3.3-70B-Instruct-Turbo | Cheaper, same abilities | −20 % | +181 ms | $0.1 | $0.32 | 131,072 | 2,169 ms | 9.2 |
| google/gemini-2.5-flash-lite | Cheaper, same abilities | −20 % | −1,457 ms | $0.1 | $0.4 | 1,048,576 | 531 ms | 90.6 |
| anthropic/claude-haiku-5-5 | Larger context | −20 % | −975 ms | $0.1 | $0.5 | 1,000,000 | 1,013 ms | 749.2 |
Request options
| Setting | Value |
|---|---|
| Returns | text |
| Request parameters | temperature, top_p, max_tokens, max_completion_tokens, stop, tools, tool_choice, response_format, n, seed, presence_penalty, frequency_penalty, logprobs, top_logprobs, logit_bias, parallel_tool_calls, reasoning_effort, audio, functions, function_call, prompt_cache_key, safety_identifier, metadata, prediction, web_search_options, verbosity, modalities, top_k, min_p, top_a, repetition_penalty |
What changed
- llama-3.3-70b-instruct added
In short
- One API posts llama-3.3-70b-instruct at 75 % below Meta's list price: $0.125 in and $0.4 out per million tokens, against $1.04 and $1.04.
- By input price it is the 10th lowest of the 55 chat models we could call on 22 November 2026.
- Its context window of 131,072 tokens holds about 262 pages of text at 500 tokens a page.
How this page is made
Prices, context sizes and abilities are read from One API's public price feed and its model endpoints each time this page is built. The time to the first token, the output speed and the price of a call are measured by us: 3 calls for each model through a One API key on 22 November 2026, entering at not recorded. A model that did not answer has no page. How we measure.
Data checked · Last changed