adds claude-2.1 (untested)

src/proxy/anthropic.ts

@@ -42,8 +42,9 @@ const getModelsResponse = () => {
     "claude-instant-v1.1",
     "claude-instant-v1.1-100k",
     "claude-instant-v1.0",
-    "claude-2", // claude-2 is 100k by default it seems
+    "claude-2",
     "claude-2.0",
+    "claude-2.1",
   ];
 
   const models = claudeVariants.map((id) => ({

src/proxy/middleware/request/validate-context-size.ts

@@ -58,16 +58,18 @@ export const validateContextSize: RequestPreprocessor = async (req) => {
     modelMax = 100000;
   } else if (model.match(/^claude-(?:instant-)?v1(?:\.\d)?$/)) {
     modelMax = 9000;
-  } else if (model.match(/^claude-2/)) {
+  } else if (model.match(/^claude-2\.0/)) {
     modelMax = 100000;
+  } else if (model.match(/^claude-2/)) {
+    modelMax = 200000;
   } else if (model.match(/^text-bison-\d{3}$/)) {
     modelMax = BISON_MAX_CONTEXT;
   } else if (model.match(/^anthropic\.claude/)) {
     // Not sure if AWS Claude has the same context limit as Anthropic Claude.
     modelMax = 100000;
   } else {
-    req.log.warn({ model }, "Unknown model, using 100k token limit.");
-    modelMax = 100000;
+    req.log.warn({ model }, "Unknown model, using 200k token limit.");
+    modelMax = 200000;
   }
 
   const finalMax = Math.min(proxyMax, modelMax);

src/shared/key-management/anthropic/provider.ts

@@ -11,7 +11,8 @@ export type AnthropicModel =
   | "claude-instant-v1-100k"
   | "claude-v1"
   | "claude-v1-100k"
-  | "claude-2";
+  | "claude-2"
+  | "claude-2.1";
 
 export type AnthropicKeyUpdate = Omit<
   Partial<AnthropicKey>,

src/shared/stats.ts

@@ -37,6 +37,6 @@ export function prettyTokens(tokens: number): string {
   } else if (absTokens < 1000000000) {
     return (tokens / 1000000).toFixed(2) + "m";
   } else {
-    return (tokens / 1000000000).toFixed(2) + "b";
+    return (tokens / 1000000000).toFixed(3) + "b";
   }
 }