GLM 5.2 Price Jumps, Claude Goes Local in India This Week

Z.ai's GLM 5.2 got more expensive to run today, and Anthropic just made Claude cheaper to buy in India. Two pricing moves, one week, pulling in opposite directions for teams choosing where to spend their AI budget.

GLM 5.2's completion price more than doubles

The token-pricing tracker The Token Ledger flagged GLM 5.2 as the single most cost-impacting change of the day, and the numbers back that up. Z.ai raised GLM 5.2's prompt price from $0.42 to $0.93 per million tokens, a jump of $0.51. Completion pricing moved further: from $1.32 to $3.00 per million tokens, up $1.68 — more than double the prior rate. The tracker's own framing was blunt: this is "the most cost-impacting change today," and it hits hardest for "anyone generating long completions," where the cost impact is described as the largest of the day.

That distinction matters for how teams actually get billed. A prompt-heavy workload — long context, short answers — absorbs the smaller prompt-side increase. A completion-heavy workload — long-form generation, code output, chained reasoning — absorbs the larger one. Teams running GLM 5.2 in production should check which side of that split their traffic falls on before assuming the price change is a rounding error.

Kimi K2.7 Code barely moved

Not every model repriced this week. MoonshotAI's Kimi K2.7 Code saw its prompt price shift by less than a tenth of a cent — from $0.72 to effectively $0.72 per million tokens, a delta of just -$0.001. For teams currently splitting inference between GLM 5.2 and Kimi K2.7 Code, that stability just became a bigger part of the argument for shifting completion-heavy traffic toward the model that didn't move.

Model	Price type	Old (/1M tokens)	New (/1M tokens)	Change
Z.ai: GLM 5.2	Prompt	$0.42	$0.93	+$0.51
Z.ai: GLM 5.2	Completion	$1.32	$3.00	+$1.68
MoonshotAI: Kimi K2.7 Code	Prompt	$0.72	$0.72	-$0.001

Laid out side by side, the gap is easy to act on: GLM 5.2's completion price moved more than three times as much as its own prompt price, while Kimi K2.7 Code's prompt price barely moved at all.

Anthropic goes local: rupee pricing in its second-biggest market

Separately, Anthropic began rolling out Indian rupee-denominated subscription plans for Claude, according to TechCrunch. The report frames India as Anthropic's biggest market after the US — a signal that localized pricing there isn't a minor regional experiment but a move in one of the company's largest user bases. Billing in local currency removes the foreign-exchange fee and card-conversion friction that previously sat on top of every subscription charge for Indian users, even before touching sticker price.

What this means if you're choosing a provider this month

Neither move should change a stable production setup overnight, but both are worth factoring into the next cost review. If GLM 5.2 is your default for long completions, the doubled completion price closes some of the gap that made it attractive against pricier alternatives — worth rerunning the math rather than assuming last month's comparison still holds. If your team pays for Claude out of India, local-currency billing is a real, if modest, cost change worth confirming on your next invoice rather than assuming it's identical to the dollar plan you signed up for.

The broader pattern worth watching: token pricing is moving in both directions at once right now, not just up. That makes locking into a single provider on cost alone riskier than it looked a quarter ago — the cheapest option this month isn't guaranteed to hold that position next month. It also means the comparison work itself has a shelf life: a cost analysis run against this week's price sheet can be stale within weeks if a provider reprices again, so treat any fixed cost comparison as a snapshot, not a permanent ranking.

For teams already metering usage per model, this is also a reminder to break out completion tokens from prompt tokens in that metering, rather than tracking a single blended rate — GLM 5.2's own price sheet just made the case for why the two move independently.

What to watch next

Keep an eye on whether other providers follow Z.ai's completion-price move, and whether Anthropic extends local-currency billing beyond India to other large non-US markets. Both would be the next signal of where AI infrastructure spending is actually heading in the second half of 2026.