Claude 4.8 Opus
currentClaude 4.8 Opus is Anthropic's high-end Opus model for complex coding, enterprise workflows, and autonomous agents. It offers a one-million-token context, adaptive thinking, and strong long-horizon results at half Fable 5's token price, though Sonnet 5 is cheaper for routine work.
Context window
1M tokens
Maximum output
128K tokens
Input price
$5 / 1M
Output price
$25 / 1M
Best for
- Complex coding and repository-wide changes
- Long-horizon autonomous agents
- Enterprise analysis with large document sets
- Work requiring strong judgment without Fable pricing
Not ideal for
- Routine workloads Sonnet 5 can solve
- High-volume, latency-sensitive chat
- Self-hosting or fine-tuning
- Native audio or video generation
PromptHQ 100-word challenge
We give every model the same miniature science-fiction mystery. The response is short enough to read in one chat bubble, while the premise tests creative writing, compression, foreshadowing, causal coherence, and instruction following.
Claude 4.8 Opus102 wordsWrite a science-fiction mystery in exactly 100 words.
A three-person research station on Europa receives a message from itself, timestamped 11 minutes in the future:
"Do not open the greenhouse door."
The door is opened anyway.
Plant two subtle clues. Give the warning a logical explanation without aliens, time travel, hallucinations, or simulations. The final sentence must change what the reader thought the warning meant.
Output only the story. Do not explain it.
The message arrived stamped 11:47 — eleven minutes ahead of our clocks. Do not open the greenhouse door. Vasquez laughed. Chen didn't answer; Chen never answered anymore. We opened it anyway.
Inside: frost, dead seedlings, the relay blinking where nothing should. Chen had rigged that unit days ago, bouncing our transmissions off the Jupiter satellite — a loop returning our own words late, stamped by its faster clock.
Chen lay curled among the planters, around the oxygen he'd siphoned to outlast our failing recyclers.
He hadn't warned us away from something inside.
He'd begged us not to let his last air out.
Run the same prompt yourself
Open Claude 4.8 Opus in Multichat with the evaluation prompt prefilled.
What the response revealed
The response contains 102 whitespace-delimited words, two over the required count.
Apart from the word count, it outputs only the requested story and avoids the prohibited devices.
Chen's relay and its faster clock explain the apparent future message, while his oxygen refuge explains the warning.
Chen's silence and the unexpected relay prepare both his location and the technical explanation.
The final two lines transform the warning into a plea not to vent Chen's last air.
The clipped sentences build tension effectively and deliver an emotionally legible reversal.
The model receives the prompt without web access or external tools. Claude 4.8 Opus is run through Claude CLI at medium reasoning effort. We preserve the response as generated apart from display rendering.
Performance
Claude 4.8 Opus benchmarks
Benchmark scores are sensitive to reasoning effort, harness, tools, token budget, prompt format, sampling, and evaluation date. Scores here retain their source and should not be treated as directly interchangeable unless the underlying setup matches.
SWE-Marathon
26%
long-horizon software engineering · Anthropic3
Long-horizon agent tasks, pass@1
59.9%
multi-step agents · Anthropic3
Long-horizon agent tasks, pass@3
67.6%
multi-step agents · Anthropic3
SWE-Bench Multimodal
38.4%
visual software engineering · Anthropic3
Family position
Claude 4 positioning
Opus 4.8 is Anthropic's high-end Opus-tier model. Fable 5 now sits above it for maximum capability, while Sonnet 5 offers a lower-cost route to near-Opus agent performance on some tasks.
Claude Fable 5
Highest general capability
$10 input
$50 output
Claude Opus 4.8
Complex coding and enterprise
$5 input
$25 output
Claude Sonnet 5
Balanced speed and intelligence
$2 input
$10 output
API pricing
Per million text tokens
Input
$5
Cached input
$0.50
Cache write
$6.25
Output
$25
Where Opus 4.8 stands out
Built for demanding work
Complex coding and repository-wide changes are central to the model's positioning, rather than an incidental capability.
Long-context capacity
The published context window is 1,000,000 tokens, making the model a candidate for large documents, repositories, and sustained agent state.
Reasoning and tools
Reasoning is supported with low, medium, high, xhigh, max provider setting(s), and the model can participate in tool-using workflows through its available API surface.
Limitations to know
Benchmarks are configuration-sensitive
Scores can move substantially with the harness, tool access, effort setting, token budget, and evaluator. Treat the table as evidence, not a universal ranking.
Context size is not guaranteed recall
A large advertised window does not mean every detail is retrieved reliably at maximum length. Validate representative long-context workloads before deployment.
Product access differs from model capability
PromptHQ and gateway limits may expose fewer modalities, tools, or tokens than the provider's first-party API.
Capabilities and specifications
Knowledge cutoff
Jan 1, 2026
Inputs
text, image, PDF
Reasoning
Supported
Default effort
medium
Supported API features
Frequently Asked Questions
Sources
- 1Claude models overview
Anthropic · official documentation
- 2What's new in Claude Opus 4.8
Anthropic · official documentation
- 3Claude Opus 4.8 System Card
Anthropic · official system card
- 4PromptHQ model registry
PromptHQ · internal product configuration