Opus 5: Near-Fable Intelligence at Half the Cost
Claude Opus 5 is Anthropic’s latest large language model, released as a near-frontier alternative to Claude Fable 5 that delivers comparable intelligence for coding and long-horizon knowledge work while charging roughly half the token price of Fable, marking a deliberate shift from premium-only AI toward cost-competitive, everyday deployment. Anthropic released its Opus 5 AI model on Friday, explicitly claiming it “comes close to the frontier intelligence of Claude Fable 5 at half the price.” That is not a minor tweak; it is a direct challenge to the idea that the most capable models must be financially out of reach for routine workloads. Instead of keeping near-frontier capability locked behind the expensive Fable 5 tier, Anthropic is now putting a high-performing model squarely in the default, affordable slot—and that tells us more about market pressure than it does about pure technical progress.
Claude Opus 5 Pricing: The Numbers That Really Matter
Anthropic’s new pricing makes the strategic pivot obvious. Fable 5 charges USD 10 (approx. RM46) per million input tokens and USD 50 (approx. RM230) per million output tokens, while Opus 5 comes in at USD 5 (approx. RM23) and USD 25 (approx. RM115) respectively. In other words, Claude Opus 5 pricing is exactly half of Fable’s sticker rate. It also lands squarely beside rival GPT-5.6 Sol, which sells for USD 5 (approx. RM23) per million input tokens and USD 30 (approx. RM138) per million output tokens. On task-level economics, the gap is just as clear: the weighted average cost per Intelligence Index task is USD 2.75 (approx. RM13) for Fable, USD 2.03 (approx. RM9) for Opus 5 at max effort, USD 1.04 (approx. RM5) for GPT-5.6 Sol, and USD 0.95 (approx. RM4) for Kimi K3. In a market where enterprises are “beset by rising token costs,” these differences stop being academic and start driving procurement decisions.
| Model | Input price (USD / MTok) | Output price (USD / MTok) |
|---|---|---|
| Claude Fable 5 | 10 (approx. RM46) | 50 (approx. RM230) |
| Claude Opus 5 | 5 (approx. RM23) | 25 (approx. RM115) |
| GPT-5.6 Sol | 5 (approx. RM23) | 30 (approx. RM138) |
| Kimi K3 | 3 (approx. RM14) | 15 (approx. RM69) |
Fable 5 vs Opus 5: Capability Gap Narrows, Economic Gap Widens
For Anthropic’s strategy to make sense, Opus 5 cannot be a budget knock-off; it has to challenge Fable 5 on quality. On the company’s own Intelligence Index, Opus 5 edges ahead with a score of 61, one point higher than Fable. Anthropic says Opus 5 is “substantially stronger than Claude Opus 4.8 across the board” and in some cases comparable to or ahead of Claude Fable 5. Another source notes Opus 5 outperformed Fable 5 on most internal benchmarks, including coding and knowledge work. At the same time, the firm is candid that Opus 5 is not designed to lead in high‑risk areas such as advanced cybersecurity and offensive biological research, unlike Mythos 5. That trade-off is telling: Anthropic is effectively reserving the highest-risk capabilities for tightly controlled models while letting near-frontier intelligence for everyday software engineering and long-horizon knowledge work spill into a far cheaper tier.
Why Anthropic Is Changing Course Now
This pricing reset did not happen in a vacuum. Fable 5 may be “arguably the most influential LLM to launch in recent memory,” but its short life so far has been turbulent. It launched on June 9, went dark internationally three days later under export controls, reappeared July 1, and by July 22 was tied up in a White House accusation that a foreign lab had covertly distilled it. That kind of scrutiny makes selling a premium, frontier-leaning model as the everyday default hard to sustain. Meanwhile, US AI vendors are facing “growing competition from lower-cost Chinese models such as Kimi K3 and GLM 5.2, making pricing just as important as performance.” In that context, releasing a model that beats Fable on most benchmarks at half the price is less a generous gesture than a defensive move: Anthropic is protecting share against cheaper rivals while repositioning Fable and Mythos as specialized, higher-risk offerings, rather than the workhorse engines of enterprise AI.
Practical Impact: Enterprises Win, Power Users Need to Adapt
For ordinary users, Claude Opus 5’s main impact is blunt: more intelligence per dollar and fewer compliance headaches around a frontier model. Enterprises that were “beset by rising token costs” can now move many coding and knowledge workloads to a model that offers near-Fable performance for significantly less. Opus 5 is billed as “the strongest balance of capability and affordability,” and is now the default for top subscription tiers, which reinforces that positioning. There are trade-offs, though. Anthropic warns that Opus 5’s default responses run longer than prior Opus models, which means the lower per-token price can be partially offset by higher token usage. Users who care about Claude Opus 5 pricing in real workflows will need to actively control output length in prompts “to control response length” instead of trusting the defaults. Still, for developers, startups and enterprises trying to keep cloud spend in check, the shift from Fable-first to Opus-first marks a clear victory: the new center of gravity in Anthropic’s lineup is no longer the most expensive model.
