The 2024 Sonnet that set the standard for practical coding work — now well behind the current Claude line at identical pricing.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
If you are choosing a model today, this page will probably talk you out of Claude 3.5 Sonnet. That is not a criticism of the model. It is arithmetic.
Claude 3.5 Sonnet arrived in October 2024 and became the model a lot of engineering teams first trusted with real code — not autocomplete, but actual edits to files that shipped. It earned that reputation. The problem is what it costs now relative to what has replaced it.
Context is 200,000 tokens. Maximum output is 8,192 tokens.
That second figure is the one that bites. Eight thousand tokens is roughly a long file — perhaps six hundred lines with comments. Ask for a full test suite, a migration script, or a rewritten module and it will stop mid-thought. You end up chunking the request and stitching the pieces, which is exactly the work you hoped to avoid.
The large context with the small output is a lopsided shape: it can read your repository but cannot write much back.
Claude 3.5 Sonnet runs around $3.00 per million input tokens and $15.00 per million output tokens.
That is the same rate the current Sonnet generation carries. Same price, roughly five times the context, and an output ceiling measured in six figures rather than four. There is no budget argument for staying, which is unusual — normally the older model is at least cheaper.
Two situations, both legitimate.
The first is inertia with evidence. If you have prompts tuned against this model, evaluation suites built around its behaviour, and output you have learned to trust, that accumulated knowledge has real value. Migrating means re-validating all of it.
The second is pinned reproducibility. Some teams need a fixed model snapshot for audit or regression reasons and cannot accept behavioural drift, whatever the improvement.
Outside those, the case is thin.
Claude 3.5 Sonnet no longer appears in Anthropic's current model line-up, and it is absent from the legacy table as well. Treat availability as something to confirm rather than assume — particularly if you are building something with a multi-year life, where a retirement notice mid-project is an expensive surprise.
| Field | Value |
|---|---|
| Provider | Anthropic |
| Context window | 200,000 tokens |
| Maximum output | 8,192 tokens |
| Input | text |
| Output | text |
| Indicative pricing | $3.00 in / $15.00 out per million tokens |
| Weights | Closed |
| Released | 22 October 2024 (v2) |
Pricing and availability move, and this generation is old enough that both deserve a check against your own account before you plan around them.
Browse current alternatives in the AI model directory, or look at Gemini 2.5 Pro if the constraint you are trying to escape is that output ceiling.
Use the recorded facts as a shortlist, then validate the model against representative inputs and production constraints.
Claude 3.5 Sonnet is categorized for Image-text-to-text, Image-to-text, Text generation, Text-to-text AI models. Its current record accepts text, image and produces text. Confirm file formats, preprocessing, and provider-specific request schemas before implementation.
The directory records 200K context and 8.2K maximum output. Listed token prices are $3.00 input and $15.00 output per 1M tokens. Treat missing values as unknown and recheck current commercial terms.
Recorded output speed is 78 t/s and time to first token is 0.60s. Hosting route, region, prompt length, concurrency, and provider load can materially change both measurements.
This page separates sourced model facts from incomplete fields. Benchmark evidence is displayed only when its source is linked and verified. Before choosing Claude 3.5 Sonnet, test task quality, tool reliability, safety behavior, data controls, rate limits, and total cost on the exact route you intend to use.
Each score below comes from a published result we could trace back to its source. Read them as evidence about the specific task a benchmark measures — a model that leads on one can easily trail on another.
Related approved models in Text-to-text AI models. Compare specifications and verify fit for your workload.
OpenAI
OpenAI's omni-era workhorse — text and image in, text out, with a cached-input rate that halves the cost of repeated context.
Google's long-context multimodal model with up to 2M token windows.
Meta
Meta's largest open-weights model, competitive with frontier closed models.
Xai via Requesty
A frontier multimodal model optimized specifically for high-performance agentic tool calling.
DeepSeek
Official DeepSeek V4 Flash release with enhanced agentic capabilities and integrated DSpark speculative decoding